Researchers from the University of Wisconsin-Madison and New York University investigated why the problems with scientific articles that were retracted — meaning they were withdrawn after publication due to the discovery of errors or some form of misconduct — went unnoticed during peer review, a process in which experts assess the validity and relevance of a study before the manuscript is accepted for publication. In an article published in the Journal of Informetrics, the authors examined 260 comments made by reviewers from a sample of 160 scientific articles that would later be retracted due to inconsistencies, plagiarism, or fraud. The retracted papers were obtained from the Retraction Watch website, while comments about them were shared by the reviewers themselves, anonymously, on a website called Publons, now owned by Clarivate Analytics.
Of the 260 reviews associated with the retracted articles, 128 recommended the studies be accepted for publication or requested minor corrections to the manuscript, equivalent to 49.2% of the total. Another 111 (42.7%) requested significant changes and only 21 (8.1%) suggested that the manuscripts be rejected — opinions that were ultimately not taken into account by the journal editors. The study suggests that journal editors and those tasked with consolidating peer-review comments pay more attention to red flags and conduct additional inspections to follow up on potential issues.
Regarding the specific problems that led to retraction of the articles, 73.8% of reviews failed to detect them, while 26.2% identified some evidence. According to the results, the peer-review process was more effective at identifying causes of retraction related to data, methods, and results than those related to plagiarism, authorship, and bibliographical issues.
The study also sought to identify the characteristics of the most perceptive reviewers who were best at detecting problems. Those most capable at pointing out inconsistencies were more experienced as reviewers (with an average of 6.19 years performing the duty) than others (4.39 years). Similarly, reviewers who were more familiar with the topic of the article demonstrated a greater ability to identify problems. According to the authors, choosing reviewers with expert knowledge of a manuscript’s research topic is crucial to detecting errors or misconduct that could lead to subsequent retraction.
Republish