Plagiarism Is Ubiquitous
2016; Lippincott Williams & Wilkins; Volume: 122; Issue: 6 Linguagem: Inglês
10.1213/ane.0000000000001344
ISSN1526-7598
Autores Tópico(s)Academic Writing and Publishing
ResumoMy decision letter to the corresponding author began gently. “Anesthesia & Analgesia has instituted 100% screening of manuscripts for plagiarism.” I explained that “our plagiarism detection software, Cross Check™a, has identified passages of text as having been taken verbatim” from a previously published manuscript. To the decision letter I attached a copy of the submission to Anesthesia & Analgesia with the plagiarized text highlighted and a copy of the manuscript that was plagiarized with the same text highlighted. I also attached the “Similarity Report” from Cross Check™ showing multiple blocks of text copied verbatim from previously published papers. All but one paragraph represented “self-plagiarism” by the senior author. Finally, I attached my editorial, “You Will Be Caught,” describing Anesthesia & Analgesia’s plagiarism policies.1 What made this decision letter memorable was that I was the senior author. Because decision letters are implicitly directed to all authors, I had just sent myself a decision letter identifying plagiarism in one of my submissions. I’m in good company. Since instituting 100% screening for plagiarism in 2010, I have sent letters identifying plagiarism to Department Chairs, Medical School Deans, Editors-in-Chief of other journals, and approximately half of the members of the Anesthesia & Analgesia editorial board. Plagiarism is ubiquitous. Plagiarism is ubiquitous because “cut and paste” is ubiquitous. We use cut and paste when we write our institutional research proposals, informed consents, grant applications, and manuscripts. We use “cut and paste” when we present our results at meetings and include our results in book chapters. I am not referring to “self-plagiarism,” an oxymoron for the common practice of using your own words in multiple papers. I am referring to real plagiarism, described by the Oxford English Dictionary as “the practice of taking someone else’s work or ideas and passing them off as one’s own.”2 If you don’t believe me, look at the last PowerPoint presentation you gave. Do you have footnotes identifying the source of every image, every graph, every table, every embedded video, and every goofy cartoon, clip art, emoji, or Wing Ding? I don’t either. In 2010 I turned to the members of WAME, the World Association of Medical Editors, for help formulating Anesthesia & Analgesia’s policies on plagiarism. I wanted to know how much verbatim duplication constituted plagiarism. A particularly judgmental editor stated that “even a phrase of duplicated text constitutes mens rea,b” evidence of a guilty mind. That’s harsh! Because this editor’s journal was published in a country from which many papers have been identified as containing plagiarized material, I screened a dozen papers from his journal. Approximately one-third had full paragraphs of plagiarized text, including a few where he was coauthor. I e-mailed these examples of plagiarism, with the plagiarized text highlighted, to the editor. He did nothing. It’s easy to be judgmental of others when speaking ex cathedra. When it comes to plagiarism, the devil is in the details. (I plagiarized that, because those aren’t my words and I haven’t attributed them to the author.c) In November 2010, I began screening every submission to Anesthesia & Analgesia for plagiarism. I explained our policy, and my screening taxonomy in a 2011 Editorial,1 the taxonomy appears in Table 1. As of today, March 6, 2016, I have screened 9550 manuscripts for plagiarism. This has taken approximately 10 hours every week for the past 5 years, an overall investment of approximately 2500 hours. I want to share with readers, authors, and journal editors what I learned from this exercise.Table 1.: Taxonomy of PlagiarismIntellectual theft is the only form of plagiarism representing serious misconduct. The ethical core of plagiarism is deception. Intellectual theft occurs when the author intentionally deceives the reader by misrepresenting the scholarly work of another author as his or her own scholarship. Judging the intent of an author is necessarily subjective. I think intent to deceive is unambiguous when an author has assembled a paper entirely by cutting and pasting from the work of others. I infer intent to deceive when a paper has multiple large blocks (e.g., paragraphs) of text copied from multiple papers. It is difficult to know how an author could repeatedly copy large blocks of text and not be aware that this was misrepresenting the scholarship of others as his or her own scholarship. When my plagiarism screening identifies intellectual theft, I reject the manuscript without further peer review. In my opinion, intellectual theft usually reflects inadequate mentorship. There may be cultural explanations as well. “Thou shalt not use the words of another author without attribution” does not have the universal transcendence of “Thou shalt not kill.” Not only is plagiarism not obviously “wrong” in the sense that murder is wrong, but in some cultures, using the unattributed words of an elder scholar is considered a demonstration of both respect and erudition. What looks to me like intent to deceive may, in the cultural context of the author, be a nod of respect and restatement of another’s wisdom. That doesn’t make it acceptable, but it does suggest an opportunity to explain our standards for ethical research conduct. Intellectual sloth represents my approach to preparing a lecture a few hours before I give it. Armed with PubMed and Google, I quickly bang out a couple dozen slides to cover the topic. We accept that a lecturer may refer to data and use images without providing every citation. Most lectures are not represented as the original scholarship of the presenter, and time is too short for any but the most critical references to have meaning. In failing to provide references for every image, I am not attempting to deceive my audience. That is the fundamental distinction between intellectual theft and intellectual sloth. In the latter, there is no intent to deceive. The citation standards for the published literature are more rigorous than for presentations because published manuscripts are assumed to represent the authors’ original scholarship. There is no place for plagiarism in the published literature, even if the cause is laziness or inattention to detail, with no intent to deceive the reader. Our ubiquitous use of cut and paste frequently results in plagiarized text in papers, but that does not make it acceptable. In 2011, I suggested double screening to eliminate accidental plagiarism. We ask authors to screen their papers for plagiarism before submission. After submission, I screen the paper again before sending on for peer review. If I identify intellectual sloth, I send the paper back to the author to correct the problem before the paper is forwarded to peer review. Plagiarism for Scientific English is the most difficult form of plagiarism I handle as an Editor. I find writing difficult, and English is my native language. I cannot imagine the difficulty of writing in an unfamiliar language. I know why they do it: their culture values papers published in English-language journals more than it values papers in their own language. I also know how they do it: they plagiarize the published literature to extract properly formed sentences and paragraphs. This often results in “patchwork plagiarism” where nearly every sentence is copied from a published paper. I don’t consider this misconduct, because the authors are trying to describe their own scholarship, not steal the scholarship of another person. There is no mens rea. There is no intent to deceive the reader. However, there are 2 problems. First, it meets the definition of plagiarism, so it isn’t acceptable. Second, I don’t trust the science, because the authors can’t read what they submitted. For several years after I initiated 100% manuscript screening, I allowed papers with patchwork plagiarism to undergo peer review. Reviewers could tell that the paper did not make sense, the expected result from weaving together full sentences found by the authors using Google. The result was 100% rejection. I changed my approach and now reject all papers with patchwork plagiarism on submission. However, I try to offer authors specific mentorship. My usual suggestion is that authors write the paper in their own language and then have the final paper professionally translated. If there are members of the Editorial Board who are fluent in the author’s language, then I recommend that the authors reach out to the Editor for assistance. Lastly, if there is a high-quality, English-language publication in the author’s country, then I may recommend submitting to that journal with the expectation that the editors can communicate with the author in his or her native language. Technical plagiarism almost doesn’t count as plagiarism at all. It reflects authors citing the original scholarship (i.e., using verbatim cut and pasted text along with a citation) but not properly linking the citation to the text so that the reader understands that the words are those of another author. I always send these back to the authors to fix. I mention self-plagiarism to dismiss it. Quoting our 2015 Guide for Authors: “The Journal does not view self-plagiarism as misconduct. Authors are permitted to reuse their own words, and are encouraged to do so when describing identical research methods in multiple papers.” Thresholds for “acceptable self-plagiarism” ranging from 10% to 30% have been suggested.3,4 Having screened 9550 papers for plagiarism, I reject these thresholds. Highly productive laboratories can and should use exactly the same words when describing exactly the same methods. If the methods are complex, then there may be many blocks of verbatim overlap with their previous papers. Perhaps they should say in these papers “As described in our previous work….” However, they don’t. I’m OK with that. Where is the deception? In my view, authors have better things to do than match each block of text in their methods to previously published manuscripts. I also don’t object to authors using the same words as in their previous articles in their introductions and conclusions when the goals and conclusions are similar to those of previous studies. This gets a little complex. For example, we retracted a paper because nearly the entire introduction and conclusion were identical to those from a previously published paper by the same authors.5 The problem was not that the authors had crossed a particular threshold of “self-plagiarism,” but rather that the reviewers and editors of Anesthesia & Analgesia believed the authors had intentionally deceived the journal by representing as significant original scholarship a paper that was only trivially different from a paper in peer review at Anesthesiology.6 Similarly, duplicate publication (publication of the same paper in 2 different journals) is a serious academic misconduct. The reason is not “self-plagiarism,” but rather that duplicate publication is egregious intentional misrepresentation by the author of the novelty of the scholarship. In some countries, it is common for non-English-speaking authors to assemble their manuscripts from their previously published papers. This is done because their prior paper was copyedited and is therefore in grammatically correct scientific English. These papers may have as much as 65% verbatim text copied from other papers by the same authors. They are using their own words to describe their own original scholarship. There is no intentional deception. I don’t see an ethical problem. Most manuscripts have multiple authors. It is frequently the case that only one author of the submission to Anesthesia & Analgesia is a coauthor of the paper from which the verbatim text has been taken. There is no way to determine whether the copied words were written by the common author. I don’t think this matters because every coauthor is responsible for the entire paper and thus every coauthor has the right to reuse the text. As a practical observation, in my own papers, I have no clue who wrote each sentence except for those cases (e.g., editorials) where I am the sole author. Finally, “self-plagiarism” is often entangled with questions about copyright. I find it helpful to realize that the core ethical concern about plagiarism is deception, whereas copyright is entirely a matter of commercial rights to an author’s scholarship. Anesthesia & Analgesia explored this issue in considerable detail in our discussion of plagiarism of textbook chapters.7–9 Samuelson10 points out that courts historically have defended the right of authors to their own words, regardless of copyright. Publishers likely recognize that suing their authors for reusing their own words is bad for business. Authors wishing to reuse large blocks of their previously published text should not be excessively concerned about copyright violation. Instead, they should ask whether the reuse of their words without reference to their previous work could be an intentional attempt to deceive. If so, they should properly cite their previous work. As an editor, when I identify large blocks of “self-plagiarism” in a manuscript, I consider whether the authors are engaged in deception. My answer is almost always “no.” For journal editors less tolerant of “self-plagiarism,” I would ask whether they would want their own papers to be similarly judged and if so whether their thresholds are based on thousands of hours of experience screening thousands of manuscripts. My rejection of thresholds is based on exactly that experience. PLAGIARISM IN SUBMITTED MANUSCRIPTS, 2008–2015 I have reviewed every instance in which plagiarism was discussed in an author decision letter from January 2008 (when we converted to Editorial Manager) through December 2015. During this time, plagiarism was discussed in 431 decision letters. The breakdown appears in Table 2. Table 2 also shows the number of papers and percentage of papers for which we requested a revision as well as the numbers and percentage of papers we eventually accepted.Table 2.: Outcome of Plagiarism ScreeningIntellectual theft was identified in 172 of the 431 papers, representing 40% of all instances of plagiarism. Intellectual theft was always met with a rejection. Intellectual sloth was identified in 124 of the 431 papers, representing 29% of all instances of plagiarism. Intellectual sloth was usually met with a request for a revision to remove or properly cite the plagiarized text. Plagiarism of any type had a poor prognosis. For papers containing plagiarism, the acceptance rate was significantly less than our overall acceptance rate of 22% to 25%. Interestingly, 2 papers rejected for intellectual theft were eventually accepted. In both cases, the authors contacted me with an explanation, a program to prevent plagiarism in the future, and a request to resubmit a revised paper. Both requests were approved, and the papers were eventually accepted. Figure 1 shows the trend in plagiarism from January 2008 through December 2015. The light blue area shows that plagiarism identified during peer review before formal screening was only rarely identified during peer review. The light yellow area shows the incidence of plagiarism identified after institution of our policy of 100% manuscript screening. In the first year of screening, I identified 188 papers with plagiarism. This trend declined each subsequent year.Figure 1.: Plagiarism over time.Although I suspected that plagiarism had declined steadily, I had not analyzed the data until writing this editorial. Figure 1 shows plagiarism decreasing every year since we started screening every manuscript and providing mentorship to authors of manuscripts with plagiarized text. We have not identified a single case of plagiarism in an accepted article since instituting our policy of 100% manuscript screening. I am certain that there are such cases, and we simply missed them in the screening. They will eventually be caught.1 However, I am equally certain that screening for plagiarism protected authors, and the journal, from the consequences of intentional or unintentional plagiarism. I base this, in part, on my own experience of being spared the consequences of plagiarism by our screening program. RECOMMENDATIONS FOR EDITORS Having personally screened 9550 manuscripts for plagiarism as of today, I offer the following recommendations for journal editors: When you identify plagiarism in a submitted manuscript, treat the author as you would want to be treated. Someday you, too, will likely receive such a letter. To protect both your journal and your authors, screen every manuscript for plagiarism. Had I not detected plagiarism in the paper for which I was senior author, it would have been caught after publication. Not only would the paper have been retracted for misconduct, but I likely would have had to step down as Editor-in-Chief. Screen every manuscript on submission. This is a lot of work! I chose to screen the papers myself, because I learned a lot about each paper in the process. However, screening could reasonably be assigned to an assistant, who brought problematic papers to the attention of the Editor-in-Chief. It is the responsibility of the Editor-in-Chief to communicate concerns to the authors and provide appropriate mentorship. Carefully distinguish intellectual theft from other forms of plagiarism. Intellectual theft is the only true mens rea. Although it is sometimes obvious,11 more commonly it is a judgment call. Whenever possible, ask the authors to revise the paper and remove the offending text before peer review. In my opinion, nearly all plagiarism is the result of poor mentorship. Even in cases of intellectual theft, consider that the authors may not understand that intellectual theft is considered academic misconduct and provide guidance and mentorship. Use Cross CheckTM from iThenticate.a It is more thorough than the free websites. The extraordinary sensitivity is well worth the modest cost of the program. a. Ignore single sentences identified by Cross Check. Single-sentence duplication is too common and too inconsequential to justify concern. I ignore any block of text fewer than 60 words for the same reason. For every block of verbatim text (e.g., several sentences), examine the source document. Most often the source will be previous papers by the same author. The vast majority of the time this represents perfectly acceptable use by the author of his or her own words from previously published papers. If you can’t write scientific papers in Mandarin, Japanese, or Turkish, then be compassionate with authors attempting to write papers in English. a. Patchwork plagiarism is very common in papers from authors who do not speak English. It is not academic misconduct. b. Expect the authors to reuse substantial blocks of text from their papers that have been previously published in English. They assume, probably correctly, that these papers have undergone copyediting to correct grammar, spelling, and logic. c. If the findings of the paper appear exceptional, but are poorly communicated because of the challenges of scientific English, consider helping authors edit the paper. You are an editor. I do this approximately 6 times each year. There is no threshold for the amount of text from an author’s previous papers that is acceptable. a. I have accepted papers with 65% verbatim text when it was all the authors’ own words and rejected papers with only 5% verbatim text when it was all intellectual theft. b. Sometimes authors include checklists (e.g., the CONSORT guidelines, paragraphs from the Guide for Authors) in the body of their manuscript. This can result in >80% duplicated text, which is completely inconsequential. Use the plagiarism screen to check for: a. Slicing a research study into minimal publishable units. b. Detecting use of a single assay applied in assembly-line style to the study of multiple drugs. c. Recycling previously published material as a new research study. Screen the manuscript from the submitted Microsoft Word file rather than the PDF file generated for reviewers. Many journals add large blocks of text to the PDF, including instructions for authors and line numbers. Remove the bibliography before you screen for plagiarism. The bibliography should be a nearly verbatim match to other papers that have the same references. If the duplicated text refers to an abstract by the authors, it is likely that the paper was presented at a meeting. Anesthesia & Analgesia considers this acceptable. You may occasionally wish to check a published paper for plagiarism. For example, a reader may claim that a paper in your journal is plagiarized. This is surprisingly difficult. Cross CheckTM will identify the paper as having 100% verbatim overlap with itself. Cross CheckTM can selectively ignore specific sources. However, there are many websites that collate published papers, including Google Scholar, the academic websites of the author, and rogue sites that egregiously violate copyright. It may be nearly impossible to identify these sites and limit screening of a published paper to papers published before the paper in question that might have been a source of plagiarized text. Be familiar with the guidelines for plagiarism published by the Committee on Publication Ethics.d These guidelines place considerable responsibility on journal editors to diligently follow up on findings of misconduct (e.g., intellectual theft). It is also worth knowing the guidelines from the U.S. Office of Research Integritye as well as any guidelines published by related professional societies. RECOMMENDATIONS FOR AUTHORS Screen every submission for plagiarism. Because cut and paste is ubiquitous, plagiarism is ubiquitous. Be particularly careful with multiauthored papers. Every author is responsible for the entire manuscript. Even if you have been careful about cut and paste, there is no guarantee that other authors have been as diligent. Use Cross CheckTM from iThenticate.a Ideally your department should have a license and a policy of 100% manuscript screening before submission. If you get a decision letter that identifies plagiarism, discuss it with your coauthors, put a plan in place to identify plagiarism for future papers, and write back to the editor to explain what happened and the plan you have instituted to prevent future recurrence. If you are asked to revise a paper for plagiarism, be certain to address every block of verbatim text. The editor will likely only give you one chance to fix it. The fact that plagiarism is common doesn’t make it acceptable. However, because it is common, journal editors must protect authors as well as their journals by screening for plagiarism. I recommend that journal editors accept responsibility for offering mentorship, guidance, support, and kindness when reporting findings of plagiarism to authors. I again recommend to authors that they screen every paper before submission.1 Finally, academic departments should institute a requirement that all papers submitted by department faculty be screened for plagiarism. DISCLOSURES Name: Steven L. Shafer, MD. Contribution: This author helped write the manuscript. Attestation: Steven L. Shafer approved the final version of the manuscript. RECUSE NOTE Steven L. Shafer is the Editor-in-Chief for Anesthesia & Analgesia. This manuscript was handled by James G. Bovill, Guest Editor-in-Chief, and Dr. Shafer was not involved in any way with the editorial process or decision.
Referência(s)