Characteristics of retracted articles based on retraction data from online sources through February 2019

Abstract

Purpose Although retractions are commonly considered to be negative, the fact remains that they play a positive role in the academic community. For instance, retractions help scientific enterprise perform its self-correcting function and provide lessons for future researchers; furthermore, they represent the fulfillment of social responsibilities, and they enable scientific communities to offer better monitoring services to keep problematic studies in check. This study aims to provide a thorough overview of the practice of retraction in scientific publishing from the first incident to the present. Methods We built a database using SQL Server 2016 and homemade artificial intelligence tools to extract and classify data sources including RetractionWatch, official publishers’ archives, and online communities into ready-to-analyze groups and to scan them for new data. After data cleaning, a dataset of 18,603 retractions from 1,753 (when the first retracted paper was published) to February 2019, covering 127 research fields, was established. Results Notable retraction events include the rise in retracted articles starting in 1999 and the unusual number of retractions in 2010. The Institute of Electrical and Electronics Engineers, Elsevier, and Springer account for nearly 60% of all retracted papers globally, with Institute of Electrical and Electronics Engineers contributing the most retractions, even though it is not the organization that publishes the most journals. Finally, reasons for retraction are diverse but the most common is “fake peer review”. Conclusion This study suggests that the frequency of retraction has boomed in the past 20 years, and it underscores the importance of understanding and learning from the practice of retracting scientific articles.

Keywords

• academic publishing
• retraction
• scientific publication
• self-correcting capability