Good Practices

Mapping plagiarism

BoasPraticas_232adaniel bueno

American physicist Paul Ginsparg suspected there might be cases of plagiarism in the arXiv, a repository he established in 1991 where physicists, mathematicians, and biologists post data from their research, submitting it to colleagues for analysis before it is published. To learn the extent of the problem, he and researcher Daniel Citron, both from Cornell University in the United States, analyzed 757,000 manuscripts that had appeared on the portal between 1991 and 2012, using software capable of identifying repetitions of passages that appear in more than one paper but without the author properly crediting the source. The conclusions of this effort, published in the December 2014 issue of Proceedings of the National Academy of Sciences, show that reuse of texts is more common than Ginsparg had thought.

According to the survey, one in every 16 authors who has submitted papers to the arXiv has committed self-plagiarism, which is the reuse of passages found in previous writings by the researcher who uses them in writing new papers. The survey also shows that one in every 1,000 authors has at one time or another copied the equivalent of at least a paragraph of text that was written by people, but failed to acknowledge them. It was also observed that the incidence of plagiarism varies geographically.

In partnership with the journal Science, Ginsparg and Citron mapped 57 countries whose researchers had made particularly significant contributions to the arXiv. Japan, the United States, and Germany, who figure among the nations that share the most papers with the repository, tend to plagiarize relatively infrequently. The percentage of authors who have a high probability of having plagiarized from other works was 5.6%, 4.7%, and 3.2%, respectively.

The highest rates were observed in countries like India (25.2%), Iran (15.5%) and China (10.7%). These are well above the global average, which is 3.2%. In Brazil, 8% of authors who submitted manuscripts to the arXiv during the period reviewed are strongly suspected of having committed plagiarism. In their article, the authors of the research attribute such practices to “differences in infrastructure and orientation, or incentives that emphasize the quantity of publications rather than the quality.” The results of the study also indicate that authors who copy other people’s work are seldom cited in other publications. Rogério Meneghini, scientific coordinator of the SciELO Brasil virtual library, believes that the pressure researchers feel to publish more and more may create conditions that favor misconduct. “Such pressure is strong in countries like China and Iran, and so their researchers seek out journals from other places, including Brazil, to feed that growing production,” Meneghini says.