Text Recycling: Forum discussion topic March 2013

Thanks for focusing the spotlight on this important topic. I've been "iThenticating" all revised AJPM papers for several years now, and am continually frustrated by self-plagiarism. I now have two lines that I repeat to our editors on a regular basis: “Self-plagiarism is, by its very name, plagiarism”; and “you’d think that researcher/authors with MDs and PhDs would be bright enough to know how to reword.” Levity aside, I’m pleased that COPE has created guidelines for us. Thanks!

Thanks for focusing the spotlight on this important topic. I've been "iThenticating" all revised AJPM papers for several years now, and am continually frustrated by self-plagiarism. I now have two lines that I repeat to our editors on a regular basis: “Self-plagiarism is, by its very name, plagiarism”; and “you’d think that researcher/authors with MDs and PhDs would be bright enough to know how to reword.” Levity aside, I’m pleased that COPE has created guidelines for us. Thanks!

It would be very helpful if there were some discussion of situations where text recycling occurs when some but not all authors are the same for the papers in question.

Also, the guidance should indicate that the authors are obligated to disclosure text recycling to the editors, that is, that the burden of disclosure rests with the authors.

Finally, some discussion as to when and how 'clarifying what is new in the subsequent publication versus the original publication" is appropriate would be helpful as well.

In my field (Pathology), and probably in others as well, a much bigger problem than recycled text (probably not so bad in an invited review article, in which the authors are specifically invited to discuss their previous publication/s, obviously with appropriate citation) is recycled photographs, diagrams/drawings, Tables and the like. Again, the rule we usually follow is to require permissions from the original publisher/s and, obviously, citations. I must say, however, as an author as well as an editor, that it has always annoyed me to have to get permission from a publisher to reuse my own original illustrations. Any comments?

Another type of self-plagiarism is when authors try to milk a research project to get multiple publications, each with a slightly (in some cases, very slightly) different angle. Assuming proper citations to the previously published articles, my general remedy is to apply a contribution metric to the paper: if the marginal contribution of the new paper is significant, I will ask the authors to indicate in their introduction that this is a "continuation" paper of a previous article and send it out for review; if the contribution is negligable, I will reject the submision.

This is a significant problem and growing I think........amazingly some argue with Crosscheck scores. A related issue is the author who multi-submits in spite of cautions not to - and you find more than one journal refereeing essentially the same (possibly then accepted) papers....if the journals are in a range of countries it may be difficult to spot

As editor I have the experience of working with Cross Check. Cross Check reports similarity index and use of similar words. The point is that report of similarity is not enough to judge about the intention of authors to copy-paste or plagiarism. I mean there are sometimes limited words specifically scientific words to state something in English and if one searches Cross Check will find almost all forms of words in previous papers. So use or interpretation of Cross Check report is something that should be learned by editors. Just having a similarity index of 30% for instance cannot mean anything if editors do not go through whole paper. In fact such similarity is not restricted to Methods section. For instance, there are limited forms to state Results in scientific papers.
I fear to reach somewhere to lack enough English words to state something to be absolutely different from previous papers. This is what we have to be careful and think about.

I think that one has to distinguish between original articles and review-type articles when applying the detailed section-by-section which is absolutely necessary for each analysis of similarity reports. In the former, the most important is issue is to prevent data duplication. Materials and Methods commonly show duplication, acceptable under most circumstances even if including data which might also sometimes be reported in the Results section (eg, clinical characteristics of patients used in several studies). Significant new data in the Results will always require new text for Discussion. If not, the paper was probably not novel enought to warrant publication anyway. Thus, my tolerance for self-plagiarism is highest for M&M, next highest for the Intro, low for Discussion and zero for Results. In review-type articles, especially those by non-native speaker writers, some guidance on reformulating text passages may be required, but "recycling" chunks of text is not such a sin here, in my opinion, as long as there is no hidden duplication of the entire paper.

From the viewpoint of Polymer Testing, the important area is the results. There must be significant or important new results but it is often reasonable to reproduce descriptions of test methods, materials and even parts of the introduction. If the test method used is not new it must be made very clear where it originates and why it is used..

A growing issue in my field is the practice of publishing a paper in a conference proceedings and then extending this paper for publication in a journal. There is often considerable overlap between these papers. I do not think authors are being disingenuous. They are typically being encouraged by conference organisers and editors to follow this practice. Conference proceedings are often considered to be valuable by the researchers in the field themselves (as they collect together a large cognate and current body of knowledge) but papers in conference proceedings are not considered to be valuable in research assessment exercises (as the crude view of research assessors is that conference proceedings are not subject to the same standards of peer review as journals). So, an author risks accusations of plagiarism (albeit text recycling) in order to satisfy these two masters: science and career. I personally would be very unhappy to see some of the suggestions above (e.g. retraction, publication of a correction) implemented in this kind of case. However, that said, I am uncomfortable with the proceedings-journal “double” publication practice.

A sometimes trite sentence or remark is surely acceptable, but wholesale repetition of one's own prose is frankly boring. Repetition of data, without due acknowledgement, cannot be used. Some opinions on their publication are sometimes draconian especially if there is only a suspicion of self-aggrandisement. More seriously, does the self-plagiarism amount to pure duplication? In a learned scientific publication self-plagiarism of solid data, without acknowledgement is unacceptable. In other circumstances, there is often no (or but a few) other ways of giving the clear information in prose form - this may apply to all literature, of course.

It is much easier to acknowledge the problem than to know how editorial discretion is to be exercised. It would probably be wise to revise the text of the document in light of the various comments submitted, noting in more detail the kinds of issues/factors that might lead to different actions being taken (there is surely no "one size fits all" solution). Greater author transparency as to the degree of overlap when a previous article is used would also be helpful. I do, however, wish to protest against the language of "self-plagiarism" -- the point is not that one is represent others' ideas as one's own, but that one is recycling one's own work. That is a different fault.

A common scenario is this: one presents a paper at a foreign university, then is asked to
publish it in their local journal or collection of working papers.
As an instance, I have several papers in local publications of Japanese universities (some
in Japanese); the same is true for Brazil. Not many people I know have access to these
sources and I regularly republish such papers (translated if needed) in more internationally
accessible journals.
I feel OK with this practice, esp. as I always refer to the original place of publication,
thus giving my foreign colleagues a small leg-up.

It has been observed in our journal, that authors conduct one study and the results are divided in two portions and submitted with a small difference in the results achieved. In such a case there is overlapping in nearly all sections. It actually falls in the section of Salami slicing with self plagiarism or Text Recycling. Our editorial board members consider this unethical and following the guidelines of COPE we ask for an explanation. Usually the answer is not convincing and we do not process the article. It is considered rejected on the grounds of ethical misconduct.

I agree with the comments above about the location of the self plagiarism within the paper being critically important eg from M&M - sometimes unavoidable, to results - unaceptable.
My other concern though is how the plagiarism software actually work. I was dumbfounded to discover that the package provided by my journal, which I only rarely use, identified the references in the bibliography as "plagiarised". When I discounted this, a manuscript with a very suspicious score initially had quite benign levels of apparent "plagiarism", requiring no further action.

This is a more subtle problem than plagiarism from other sources. The proposed Guidelines do strike the right note. My only suggestion is that it would be helpful to operationalise the Guidelines by saying that, say 10% duplication in Introduction and Methods sections (but not Results) should not normally be cause for comment.

BWT, (to Chris Barrow), based on my use of other, similar systems, I wouldn't rely solely on Crosscheck scores, but also eyeball the extent of the actual match. Specious matches can occur, e.g. including a sequence found in a database.

The guideline might state when text *should* be recycled. Text should be recycled when a paper is reporting different findings from a study previously reported. Why should synonyms be used for methods? If the methods were identical they should be described using the same words. This might apply to papers reporting subgroup analyses from clinical trials or observational analyses of trial data. Similarly, when describing participant characteristics in those cases, hopefully they are the same, if the sample was the same. Using synonyms and mixing up words serves no purpose in those cases. On the other hand, there is clearly duplication that should not happen, like reporting the same discussion, or analytic results. Introductions could present similar arguments but would be unlikely to be justifiable as identical.

The use of such term as 'self-plagiarism' makes false impression that 'text recycling' is a kind of plagiarism. This is not the case. It is legal under some circumstances to insert a section of text from a previous article into a new one. The problem is that authors do not inform editors about the 'recycled' sections. I suggest to avoid the term 'self-plagiarism' and to add the phrase, "Editors may ask authors to confirm that there are no recycled sections in the submitted manuscript or to provide the list of such sections".

In the humanities and the social sciences, where book publications have had, and still enjoy, prominence, the dynamics of recycling are particularly to be seen 1) in the interaction between book and paper/article publication, with books chapters redesigned as papers/articles and vice versa, and 2) in the effort to broadcast results originally relating to language/culture contexts different from English to a larger audience by way of translation into English. As long as the publication history is indicated clearly, both to editors and readers, there should be no problem in this.

Self plagiarism is an oxymoron. I work in systematic reviews and I would prefer to see the same correct methods used across a number of SRs than in correct ones. There are not many ways to describe how bias was identified for example and I am involved in 10-20 new reviews a year. Are some of the commenters to this list wanting every one of these to be totally different??
Obviously this doesnt apply to results etc

I don't see what's wrong with authors writing "The methods used in this study were previously described xxx (article reference)" thereby indicating who was the originator of the text and freeing up publication space for genuinely original work. Such an approach would prevent the not uncommon situation where the authors of several papers containing recycled text are not all the same people, or their names appear in a different order, or the corresponding author is different from paper to paper. On one occasion in my experience, quite large sections of text were being passed around between a number of people and papers and it became unclear (certainly to us editors, but also, I think to the authors) whose text it actually was in the first place. It was beginning to blur into genuine plagiarism, except that we could not untangle the paper trail.

I wondered whether are any implications in respect of publications arising from doctoral theses, which are increasingly being archived in open access repositories by universities. This may well be a wider issue, but text recycling would seem to be unavoidable in such cases, and perhaps worthy of some mention or specific caveat in published guidelines?

Some medical journals like JAMA and NEJM will not accept articles with previously published information. Dr. Arnold Relman, NEJM editor explained his readers wanted new, fresh information, not a rehash. I agree, and recommend searching submissions for previous publication of text as well as photos, if only to avoid copyrite problems. A related topic is when a single research study is conducted by four individual co-authors, who then republish results with slight changes while changing the title and rotating the names of first/last author. When the Journal of Prosthetic Dentistry began in 1951, the average number of authors for each article averaged about 1.5 names during the first decade, and increased gradually every decade, reaching 4+ in the 1990s. PubMed uses only four name.

We routinely run submitted manuscripts through iThenticate software to detect issues related to plagiarism. Typically, the iThenticate report indicates a “similarity” level of less than 20%. We do have occasional papers that are in the 30-40% similarity range. The similarity is mostly associated with duplication of methods regularly used by the authors’ laboratory. We recently received a manuscript from a reputable laboratory (indeed, from a lab that publishes regularly in our Journal) that came through iThenticate with a similarity score of 61%. When we went through the report to understand the details, it was clear that the problem was again due primarily to repetition of methods – i.e., “self-duplication.” While we have no desire to force authors to “reinvent the wheel” every time they write a paper, we are concerned that authors are – more and more – establishing a template for their reports so that they can simply “plug in” the latest variable. In this case, the authors are reporting the effects of a new drug, using measures that the lab uses – and has published on – routinely. They had previously reported (in our Journal) on a different drug – but using the same procedures. What is the proper balance between asking the authors to provide “new” text for their manuscripts and not imposing an unnecessary burden on them?

New text versus old is an issue, but one can paraphrase (it is a skill and potentially more difficult for non-native english speakers/authors). Beyond that surely if one is doing something new, then the "angle" of the introduction should change over the years? However, in some fields gathering observations on a particular system over time across many papers is an essential exercise. Perhaps in such instances it is the journals that are at fault: what would be wrong with a short one paragraph introduction? This would help. Indeed most of the PhD theses coming out from my lab now have a short ~10 page introduction, rather than the UK norm of 35-50 pages and no one seems to suffer. Indeed, this has been welcomed by many (but not all) of the external examiners.
Highlighted by others here is the question of data re-use. I have recent experience of brining issues of data re-use to the attention of various journals (openly). It is clear that the problem is dealt with piecemeal. It would seem that serial offenders may end up with serial corrections from different journals, each obtaining retrospective copyright transfer from the original sources. However, the problem of serial data re-use is not dealt with and the offenders may continue. So some system is required to put a barrier in place for serial offenders. Retraction seems to be the best deterrent. This, after all is the medicine the institutions who employ many authors, Universities, prescribe to their students and it would be perverse if there was one law for students and another for their teachers.

I'd be interested to know what the justification might be for demanding a new and original introduction every time. If a lab is pursuing a drawn-out project, it's likely to publish several papers on the topic before it's through (at least in chemistry, my field). The justification for pursuing the research isn't likely to change over the course of the project, and the state of the art may not change significantly either (with the exception of the papers previously published, which ought to be cited, obviously). Given that, why should an author bother finding new ways to explain why their research has merit every time they submit a new paper? Why not develop a cogent, concise explanation of the relevant background, and use it every time, updating as necessary?

How about educating people about copyright and ethical practices in publishing? One also must take note of the fact not everybody (different cultures) share the same notion on plagiarism (self or otherwise)- the western concept of "owning the words". . My experience with students from different cultures (Asian, East Asian) reveal that most of them are appalled at accusation of plagiarism or duplicate publication. They dont seem to do it intentionally. Many research papers have addressed this issue. This Forum should take a more holistic approach of educating people. I strongly feel that self-plagiarism is misnomer or a oxymoron term. Better to say copyright violation.

Regarding the retraction conditions I think it is somewhat terminating to consider joint products or by products in research and publications. So if the same process or almost identical process could lead to another conclusion why should that not be publishable as a separate paper by the publishers. It is the scientific contribution and the new knowledge that is created which is of essence. The reason I am saying this is because as many of us know perhaps all of us involved in research know and especially those coming from non English mother tongue countries know very well that to write a paper involving appropriate language takes a long time and effort which may be avoided if we follow a system of modular transformation that is using existing papers word processed as a process which can be changed in certain modules to give rise to another research output, this works for manufacturing processes, I don't see what is wrong with applying such an approach to research output, infact I think publishers and the academic community especially Repec should encourage such a process.

I strongly agree with two points previously made:
1. Text recycling is a much better term than self-plagiarasm, which is an oxymoron. In addition, it's great to focus the policy on text recycling per se as distinct from copyright violation.
2. Richard Saitz's statement, which I reproduce here: "The guideline might state when text *should* be recycled. Text should be recycled when a paper is reporting different findings from a study previously reported. Why should synonyms be used for methods? If the methods were identical they should be described using the same words. This might apply to papers reporting subgroup analyses from clinical trials or observational analyses of trial data. Similarly, when describing participant characteristics in those cases, hopefully they are the same, if the sample was the same. Using synonyms and mixing up words serves no purpose in those cases. On the other hand, there is clearly duplication that should not happen, like reporting the same discussion, or analytic results. Introductions could present similar arguments but would be unlikely to be justifiable as identical."

I agree to disagree vehemently that self-plagiarism is a copyright violation or self infringement. If this is considered unethical in publication ethics, then predominant number of scholars globally will be guilty of this copyright violation. The self-duplication will therefore cut across translation of research articles to different languages, culling and publishing articles from dissertations and thesis, publishing articles in book of abstract proceedings and journals etc. In systematic analysis and meta-analysis a reasonable level of parallelisms are expected for comparability, internal and external consistencies off findings especially in the study design, sampling methods, setting etc. Introduction, methods, limitations of study could present similar arguments. Title, aim or objectives and conclusion should be exclusively different.

There are several important problems here. First we need to dispose of the problem of duplicate publication of results. This is clearly wrong, and is covered by existing codes of publication ethics. But then there is the more complex question of how to describe research questions and methods in studies that lead to multiple publications. It is important, I think, that authors deal with this in a rational and standardised way. One way to do this is to publish a research protocol that describes these in detail, and then to carefully refer back or quote the protocol in subsequent results and implications papers. This works well, but is not always possible. In which case standardised descriptions of research questions and methods are desirable. There are two reasons for this. First, failure to do this interferes with research synthesis, systemic review, and meta-analysis because it may not be possible to distinguish between two different papers reporting results of the same study. In the work leading up to a systematic review, we found three papers, published several years apart, with many different authors, and describing different sub-group analyses that seemed to support a particular approach to clinical care. By accident, we discovered that these were analyses of data from the same study, and we were able to treat them as one. Had we not done so we would have concluded that a particular intervention was supported by very robust evidence from several large studies, an entirely erroneous result. Second, failing to describe questions and methods in an exact and standard way misleads other researchers who wish to replicate a specific study or group of studies.

The notion of self-plagiarism is an entirely new one. It is an artefact of the software routinely used to search of evidence that authors have copied others, but which has also revealed that some authors have copied themselves. This is wrong. But there is now a drift towards seeing the re-description of ideas as misconduct. This is just as wrong.

If from a single study group the results are published as separate two papers with the same material and methods for example results during induction (induction characteristics)and results during recovery (recovery characteristics) with different introduction ,results and discussion ,is it pliagrism