A new inquiry into what causes papers to be retracted has found that most retractions—more than 67% of those analyzed—are because of researcher misconduct, not error. Moreover, the percentage of retractions due to such misconduct has been on the rise since 1975. The study is the most comprehensive look yet at what accounts for retractions and helps provide a foundation for proposals on how to minimize retractions and misconduct (1).
For a more in-depth look, Feng and two collaborators reviewed more than 2000 papers listed as retracted in the online database PubMed. In order to determine the cause for each retraction, they went beyond journal retraction notices, which were often vague and didn’t reflect a reason for pulling the papers. So, the group delved into media coverage of the papers, reports from the Office of Research Integrity, and in one case consulted the researchers themselves.
“When we started drilling down into some of the individual papers, we found that in many cases you had to go outside the conventional retraction announcement to really find out what happened,” said Fang. The classification of retraction purpose was changed after consulting outside sources for 158 articles.
Of the 2047 papers analyzed, 889 were retracted because of fraud or suspected fraud and 490 were retracted because of plagiarism or duplicate publication, all of which were classified as researcher misconduct. By comparison, 437 were retracted because of errors. When the researchers analyzed additional data on the retractions, they found that highly cited articles, those published in high-impact journals, and those written by researchers in the United States, Germany, and Japan were more likely to be retracted because of fraud. Plagiarism and duplicate publication was more often found in retractions from lower-impact papers and countries with less established research programs.
“We realized that retraction as a phenomenon is complex,” said Fang. Many factors—including pressure on scientists as well as publication processes—likely contributed to cases of misconduct that led to retractions.
“Our goal with studying retractions was not so much to focus on these unhappy episodes as it was to have an accurate starting point to discuss retractions and what can be done to discourage research misconduct,” said Fang. “Getting the facts is the first step in having a serious conversation.”
Fang and his collaborators hope to pursue future work on how peer-review processes affect the quality of scientific publications and how different publication strategies can support good scientific research.
- Fang FC, Steen RG, and Casadevall A. Misconduct accounts for the majority of retracted scientific publications (2012). Proceedings of the National Academy of Sciences Early Edition, online October 1, 2012.