Tracing the Retraction Cascade: Identifying Non-retracted but Potentially Retractable Articles

Scientific advancement is driven by new discoveries that often build upon previous knowledge - usually in good faith. Yet recently, there is an increasing number of either erroneous publications or even cases of scientific misconduct, as evidenced by rapidly growing numbers of retracted articles. Unfortunately, the problem is not only limited to retracted articles, but also with all articles citing them. Whenever an article is retracted, it raises concerns about the validity of its findings, rendering it an unreliable source for subsequent citations. Consequently, all articles citing a retracted article come under scrutiny. However, Re-evaluating all citing articles is a tedious, unrealistic, and unnecessary process because not all citations hold equal significance. Therefore, we aim to identify citation patterns that lead to cascading retractions. We investigate around 5000 articles citing retracted articles, including 953 cases of cascading retractions. Moreover, we propose a retraction-centric approach to rank articles that are close to retraction by measuring their similarities to bibliographically coupled retracted articles. This study presents an alternative to exhaustive re-examination, offering a more efficient means to scrutinize the articles that require re-evaluation.