Below is the text of my letter to the Editor of the Journal of Informetrics, commenting on the article by Han et al. (2025):
Han et al. (2025) offer a detailed quantitative examination of Chinese medical researchers with retracted publications from 20 leading hospitals, analyzing author characteristics, retraction drivers, and career consequences. Comparing retracted researchers to matched non-retracted peers, the study uncovers distinct patterns across career stages: early-career researchers with retractions underperform academically, experience steep declines in citations, reduced collaboration, and limited career mobility, whereas senior retracted researchers maintain high productivity, influence, and expansive collaboration networks, reflecting cultural and institutional leniency. Probit regression demonstrates that output-driven incentives, such as rapid increases in publication and citation rates, raise retraction risk, while broader collaboration networks mitigate it. Institutional peer pressure exerts minimal influence. Strikingly, retractions due to scientific errors result in harsher professional penalties than those caused by misconduct, including larger citation and collaboration losses, challenging prior assumptions that fraud is most damaging. Using a difference-in-differences framework, the study quantifies these impacts, highlighting early-career researchers’ disproportionate vulnerability.
However, Han et al. could have advanced a stronger and more explicit critique of the current retraction system. The prevailing practice of treating all retractions as equivalent poses a serious challenge to both scientific integrity and fairness within the academic ecosystem. In its present form, retraction functions as a blunt instrument, failing to distinguish between fundamentally different causes and intentions underlying a paper’s withdrawal. A single undifferentiated category is used to encompass acts of deliberate misconduct—such as fabrication, falsification, or plagiarism—as well as inadvertent methodological errors, miscalculations, or oversights in data interpretation.
This lack of differentiation collapses essential distinctions between ethical violations and honest mistakes, often leading to disproportionate and unjust outcomes that unnecessarily damage the careers and reputations of researchers acting in good faith. Retractions can have profound and lasting consequences for scientific careers, extending well beyond the immediate withdrawal of the paper in question. Evidence shows that a retraction can reduce not only citations to the retracted work but also to the authors’ prior publications, generating long-term negative effects on scholarly reputation, particularly when authors do not proactively disclose or self-report the error (Lu et al., 2013). In general, retractions are associated with significant declines in citation counts, and even those resulting from honest mistakes can lead to severe reputational penalties for scientists (Azoulay et al., 2017).
Although many retractions arise from genuine errors, public perception often equates them with misconduct or fraud, creating a pervasive stigma (Behavior, 2021). This fear of reputational damage can discourage researchers from correcting the scientific record, potentially driving talented and conscientious scientists out of academia. Retractions thus function as a form of stigma that can call into question both a researcher’s professional competence and ethical integrity, irrespective of whether the underlying issue was inadvertent. Studies have highlighted that retraction notices themselves can communicate this stigma, shaping reputational outcomes within the scientific community (Xu & Hu, 2022).
Moreover, recent empirical work demonstrates that retractions often attract intense attention, which can precipitate departures from scientific careers, disproportionately affecting researchers whose work is under greater public scrutiny. In addition, the structure of citation and collaboration networks is typically altered following a retraction, further amplifying the long-term consequences for the affected authors (Memon et al., 2025). The consequences of this undifferentiated approach are multifold. At the individual level, researchers who make genuine mistakes face severe professional penalties, including sharp reductions in citations, collaborative opportunities, and career mobility, regardless of the absence of any malicious intent.
Early-career scientists are particularly vulnerable, as their networks and reputations are still nascent, making them less able to absorb the shock of a retraction. At the systemic level, the conflation of misconduct with honest error can discourage transparency and self-correction. If admitting an error is effectively indistinguishable from being labeled dishonest, researchers may avoid reporting mistakes, and journals may hesitate to issue corrective notices. The resulting culture of fear inhibits the epistemic process: science thrives on error detection and correction, yet this vital mechanism becomes stigmatized, undermining the reliability of the scientific record.
Therefore, Implementing a principled retraction typology could provide a practical and thoughtful solution to this problem. A typology would categorize retractions according to the underlying cause, differentiating between intentional misconduct, gross negligence, honest error, and procedural or editorial issues. By making the reasons for retraction explicit and transparent, this system would allow the scientific community, funding bodies, and the public to interpret the significance of a retraction accurately.
Intentional misconduct would remain clearly marked and appropriately stigmatized, preserving accountability and maintaining trust in scientific norms. Meanwhile, honest errors could be framed as part of the self-correcting nature of science, with minimal damage to the researcher’s reputation and career trajectory. Insights from legal and penal frameworks can provide valuable guidance in designing a fair and effective system for scientific retractions. Just as criminal law carefully distinguishes between intentional crimes, negligence, and accidents, academic corrections should likewise differentiate deliberate misconduct from honest mistakes.
In legal systems, penalties are proportionally scaled according to factors such as intent, the harm caused, and the foreseeability of the outcome, ensuring both fairness and accountability. For example, deliberate fraud is met with severe punishment, whereas acts of negligence or inadvertent error typically result in milder sanctions or corrective measures. Translating these principles to academic publishing suggests that a structured retraction typology could clarify distinctions between different types of retractions, providing transparency and proportionality in response to the underlying cause.
Level 1 – Malicious Intent: Fraudulent Retraction
- Fabrication: Making up data or results and presenting them as genuine.
- Falsification: Manipulating research materials, equipment, processes, or data such that the research record is misrepresented.
- Plagiarism: Appropriating another person’s ideas, processes, results, or words without proper attribution, including verbatim or near-verbatim copying that materially misleads regarding the author’s contributions. This excludes limited use of standard methodological phrasing, text recycling (self-plagiarism), or authorship/credit disputes.
Level 2 – Negligence: Negligent Retraction
- Errors resulting from inadequate methodology, oversight, or carelessness, without intent to deceive. Examples include flawed experimental design or mismanaged data that do not constitute deliberate misconduct.
Level 3 – Honest Mistake: Corrective Retraction
- Errors occurring despite reasonable care, such as miscalculations, data handling mistakes, or unforeseen methodological limitations.
- Self-plagiarism, depending on the extent of text overlap.
Regarding the inclusion of self-plagiarism as a Level 3 (honest mistake), several clarifications are warranted. Although self-plagiarism is frequently framed as an ethical violation, concerns surrounding it appear overstated. Callahan (2014) argues that self-plagiarism has been inflated into a perceived crisis despite limited evidence that it meaningfully harms the scholarly record. Editorial approaches vary widely: some journals impose rigid similarity thresholds (e.g., 10%), while others suggest that the reuse of even a sentence or paragraph may be unethical. Such absolutist interpretations overlook standard scholarly practices, including the iterative development of ideas, the reuse of methodological descriptions, and the repetition of theoretical frameworks for clarity and continuity. Crucially, unlike traditional plagiarism, self-plagiarism involves the reuse of one’s own previously published material and does not entail deception regarding authorship. Moreover, major institutional frameworks—such as MIT’s Procedures for Dealing with Misconduct in Research and Scholarship—do not classify self-plagiarism as research misconduct, underscoring its secondary ethical status. Treating self-plagiarism as a moral failing risks diverting attention from genuinely serious violations, such as fabrication or falsification, while penalizing routine academic practices. From an ethical standpoint, self-plagiarism should merit retraction only when it reaches clearly excessive levels—typically above 30% of textual overlap—and when such reuse undermines the novelty of the work or materially distorts the scholarly record. Below this threshold, concerns are more appropriately addressed through transparency, citation, or editorial correction rather than punitive measures. Overemphasis on minor textual overlap risks fostering compliance anxiety rather than promoting substantive ethical integrity.
Pacheco-Torgal 