Slaying the undead: How long does it take to kill zombie papers?

This article investigates the persistent impact of zombie papers, i.e., flawed publications—either already retracted or likely to be retracted—that continue to circulate within the academic system despite their discredited or discreditable status. Drawing on a large sample of 25,480 retracted research articles published between 1923 and 2023, we identify the factors that prolong the presence of flawed research in the scientific record before formal retraction. We show that articles retracted for serious misconduct, such as data fabrication, tend to experience longer retraction delays, particularly in contexts where detection mechanisms are weak and editorial or institutional responses are slow. Our findings also reveal significant disparities across geographic regions and disciplines, with under-resourced institutions exhibiting longer retraction delays. Journal practices — such as open-access versus subscription-based models — influence retraction dynamics as well, with subscription-based journals being more effective in implementing timely retractions. In contrast to the existing literature that has mostly documented aggregate citation trends, our study offers a dynamic, multivariate framework for understanding zombie paper survival. Our findings contribute to the growing literature on scientific integrity by revealing how systemic weaknesses in editorial governance and scientific transparency can delay the timely correction of the scientific record. Based on these results, we propose several policy recommendations to reduce zombie paper persistence—emphasizing the need for stronger data transparency requirements, investment in detection tools, and global editorial standards aimed at safeguarding the reliability of scientific knowledge by minimizing the epistemic risks posed by zombie papers.