Slaying the Undead: How Long Does It Take to Kill Zombie Papers?

Résumé 0

This article examines the persistent impact of zombie papers, i.e., retracted or destined-for-retraction publications that continue to influence academic discourse through ongoing citations despite their discredited status. Relying on a large sample of 25,480 retracted research articles over the 1923-2023 period, we introduce a novel methodological framework combining survival analysis with the innovative Zombie Population Decay Dynamics (ZPDD) model, a theoretical approach designed to simulate the long-term persistence and decay of zombie papers under various editorial interventions. We identify key factors affecting retraction timing and zombie papers' persistence. Serious misconduct, such as data fabrication, significantly delays retractions, while geographic disparities exacerbate inefficiencies, with certain regions facing prolonged processes. Journal practices, such as open-access versus subscription-based models, also influence retraction dynamics, with subscription-based journals exhibiting faster corrective actions. Developing a mathematical optimization framework derived from our ZPDD model, we determine the most effective mix of editorial policies while balancing practical feasibility and intervention intensity. The findings highlight data transparency as the most impactful intervention for reducing zombie papers' persistence, followed by enhanced plagiarism detection and reproducibility measures, such as pre-registration and replication studies. Overall, a well-balanced combination of targeted editorial interventions can substantially accelerate retraction processes and limit the detrimental influence of zombie papers on academic discourse.