American researchers created a database to serve as a reference for studies on the use of preprints, a type of prepublication scientific manuscript that presents preliminary results before they have been peer reviewed. The database is named WithdrarXiv, a combination of the words withdraw and arXiv, after one of the most renowned public preprint archives, created in 1991 for research results primarily in the fields of physics, astronomy, and mathematics. Containing approximately 14,000 papers uploaded to and subsequently removed from arXiv, it was described in an article shared on the repository itself in December by computer scientists from the University of Pennsylvania and Oregon State University. The paper also examines the reasons why authors chose to withdraw their preprints from the platform. Factual and methodological errors were behind the removal of more than 6,000 texts. Another 3,100 were removed for being incomplete (the work was completed later), while 2,800 contained data that was later incorporated into more comprehensive studies. Less frequently, preprints were removed due to legal issues, typos, plagiarism, or because they did not present any novel finding.
The study found that the profile of withdrawn preprints is very different to the profile of articles retracted from scientific journals after publication due to errors or evidence of misconduct. “Factual or methodological errors in manuscripts are generally rooted out through peer review,” said Vedran Katavić of the University of Zagreb, Croatia, who studies the retraction of academic papers but was not involved in the WithdrarXiv study, according to the journal Nature. “The impetus to publish on preprint servers is to be the first, and not necessarily to be completely correct,” he added. Information scientist Jodi Schneide of the University of Illinois Urbana-Champaign points out that while preprints are generally removed at the authors’ initiative, retractions are made by the editors of scientific journals, frequently without the authors’ consent. Schneide told Nature that the analysis of why preprints are withdrawn could help reduce the incidence of problematic manuscripts.
Republish