How do journals deal with problematic articles. Editorial response of journals to articles commented in PubPeer

Abstract

The aim of this article is to explore the editorial response of journals to research articles that may contain methodological errors or misconduct. A total of 17,244 articles commented on in PubPeer, a post-publication peer review site, were processed and classified according to several error and fraud categories. Then, the editorial response (i.e., editorial notices) to these papers were retrieved from PubPeer, Retraction Watch, and PubMed to obtain the most comprehensive picture. The results show that only 21.5% of the articles that deserve an editorial notice (i.e., honest errors, methodological flaws, publishing fraud, manipulation) were corrected by the journal. This percentage would climb to 34% for 2019 publications. This response is different between journals, but cross-sectional across all disciplines. Another interesting result is that high-impact journals suffer more from image manipulations, while plagiarism is more frequent in low-impact journals. The study concludes with the observation that the journals have to improve their response to problematic articles.