Guidelines for Plagiarism

AT THE ANNUAL MEETING IN SAN DIEGO, THE INFORMS PUBLICATIONS COMMITTEE adopted a new plagiarism policy that is intended to guide authors and editors for all INFORMS journals. This policy was the result of a summer’s worth of work by a subcommittee on plagiarism where I was assisted by Bob Fourer, Paul Gray (current editor of TutORials) and three former editor-in-chiefs: Larry Wein (Operations Research), Wally Hopp (Management Science) and Erhan Cinlar (Mathematics of Operations Research). The document went through dozens of iterations with many more iterations as it was reviewed by the EICs on the full publications committee. At times it seemed as if we were living through our own version of designing health care legislation.

Why was this so hard? After all, we all know what plagiarism means.Indeed,we had little difficulty designing the essence of plagiarism in a single sentence: “Plagiarism is the copying of ideas, text, data and other creative work (e.g. tables, figures and graphs) and presenting it as original research without proper citation.” Why did this document take an entire summer to write?

Ninety percent of the time was spent on the issue of guiding the re-use of an author’s own work. This is sometimes referred to as self-plagiarism, which seems to immediately brand re-use of your own work in the same category as the theft of words and ideas from someone else. In the academic research community, plagiarism ranks as one of the most severe crimes a researcher can commit, perhaps comparable to faking results, and one that brings harsh penalties.Yet it became quickly clear that within the academic community, there are widely varying views on proper practices for using work from our own published work. This diversity of viewpoints exists to a degree even among the EICs, and resulted in one instance of a senior faculty member being subjected to the embarrassment of a plagiarism review (which was dismissed) from following a policy that he thought was completely reasonable.

The purpose of this article is to highlight the new policy and to begin a process of bringing INFORMS toward a common policy on these matters. Every author submitting a paper to an INFORMS journal through the Manuscript Central system will have to check a box acknowledging that the paper conforms to this policy. At this point in time, there are differences of opinion on some key issues, and authors have to understand when they need to contact a journal’s editors to resolve certain questions.

It became clear that there are widely varying views on proper practices for using work from our own published work.

The most extreme instance of re-using your own work is to submit the same paper to two different journals at the same time. There is complete unanimity, across all INFORMS journals and across the academic publishing community, that this practice is not allowed. But what about copying the description of a problem or how a series of experiments were run to a subsequent paper that is building on the work of a prior paper? What if you want to copy a single section from an earlier paper that describes an algorithm which you are going to use in new research?

This question is complicated by the separate issue of copyright violations. Authors are generally given fairly wide latitude regarding the reuse of their own work, even word-for-word. But the issue of copyright violations has to be kept separate from the evaluation of what is new and original in a paper so that the editors and referees can judge the incremental contribution of a paper over prior work. Even if you are allowed to copy your own prior work, a journal needs to know when this is happening.

In many conversations over the summer, I found that a number of senior members of our community took for granted that incidental copying of prior work, such as a description of a dataset, was acceptable when it was clear that there was no intent to claim credit for the same work twice. Researchers often write a series of papers building on prior work as they work through an idea, and it can seem silly to force an author to paraphrase rather than copy from a published paper a few passages that are not central to the contribution of a subsequent paper. Of course we can expect that the author has cited the prior work, but when reviewing a paper, the editorial staff needs to know when the words of a new paper have been copied from an earlier paper.

The new guidelines limit copying from papers that are not your own to a few words or sentences that are clearly highlighted using quotes, italics or indenting, with the appropriate citation. However, authors are given more latitude using material from their own papers. The new guidelines state: “More extensive word-for-word copying of one’s own work is permitted (with permission from the holder of any copyright), but this must be clearly indicated in the article.”This can include copying several paragraphs or even an entire section without the usual devices of quotes or indenting, but the paper has to indicate when a section has been copied. For example, the guidelines suggest wording such as,“This section is taken from section x.x of (citation),” which means the section has been copied. If the substance of a section is used from a prior paper, an author might say,“This section is based on section x.x of (citation).”

The most contentious issue arose in the use of material that has been (or will be) published in the proceedings of a conference. INFORMS as a community focuses primarily on journal publications,but there are many conferences in which our members participate, including some within INFORMS, where attendance requires submission of a proceedings paper. For the plagiarism committee, we had to address the question of whether it was acceptable to publish work with a conference and then send a longer (and usually more complete) paper containing the same results to an INFORMS journal. To some, this was “self-plagiarism” while for others, presenting (and publishing) research at a conference prior to sending it to a journal is common practice.

Republishing results that have previously appeared in a conference proceedings (possibly word for word) is acceptable within IEEE (notably the IEEE Transactions on Automatic Control) and computer science (notably the machine learning journals). Within the EICs of INFORMS journals, some support this policy while others strongly oppose it. This complicates life for authors whose research activities span multiple communities. If you submit a paper to an INFORMS journal that builds on a prior conference proceedings paper, you need to not only cite the proceedings paper, but also make it clear the extent to which results have been previously published. Some editors do not mind that material was previously published in a proceedings, while other editors intend to treat proceedings as comparable to any journal article.

Complicating the issue even further is the submission of a proceedings paper to a conference that is basically a synopsis of a paper that has already been submitted to a journal (and is still under review). Some editors view this as a form of parallel publication, and insist on being notified when this is happening, since the proceedings paper may appear before the journal paper. The guidelines state, “If the original paper is still under review, the author must notify the editor of the journal reviewing the original submission and follow the policies of that journal. Failure to do so may be construed as parallel publication of a result.”

In time, journals should state their policy toward submitting proceedings papers on their Web site. Lacking this, it may be a good idea to contact the EIC before submitting a paper that is based on a proceedings paper to learn the journal’s policy. Some editors do not view proceedings papers as true publications, while others treat them as comparable to journal articles.

The plagiarism policy document also provides guidelines on the re-use of empirical data and the re-use of mathematics, each of which poses some unique issues. There is also a discussion of copyright, which is an issue separate from plagiarism. The document discusses the difficult issue of fair use, which is supposed to guide authors as to when they may or may not use copyrighted material without permission. Given the lack of precision in the doctrine of fair use, the INFORMS publications office has generously offered to help authors of INFORMS papers answer questions about whether material (by any author, from any journal) can be used without permission. You can e-mail your questions to [email protected].

Ultimately, the goal of the new policy is contained in the following statement taken from the new guidelines:

The overarching goal of this policy is transparency, so that the editorial staff understands what is new and original, and the degree to which the paper is drawing on the work of others or the authors. If you are not sure how to properly credit work that is presented elsewhere (such as a parallel publication which is also under review or a conference proceeding), the best strategy is to describe the situation in a cover letter to the editor.

The new guidelines can be viewed in their entirety (at the moment) at www.castlelab.princeton.edu/plagiarism.htm.