摘要：

【目的】学术界通过撤销问题论文来进行自我纠错。对全球撤销论文的特征进行分析,以揭示历史、现状和规律,旨在为各类学术共同体提供完善科研诚信管理的有效对策。【方法】从Retraction Watch(RW)收录的全球撤销论文中,抽取1980—2023年撤销的32868篇期刊论文作为基础数据,将撤销原因区分为“涉嫌科研失信”和“未直接表明涉嫌科研失信”2大类、12个小类,分别从撤销的时滞、原因、学科、期刊以及国家/地区等5个方面进行统计,并深入分析各个小类撤销原因的表现差异。【结果】全球撤销论文数量持续增加,涉嫌科研失信原因的撤销论文占比显著上升;近10年的撤销时滞为3年左右,且近5年呈现变短趋势;基础生命科学和健康科学是撤销论文的“重灾区”,商业/技术学科逐步成为多发学科;有国际合作的论文撤销比例较非国际合作的略低,而发展中国家涉嫌科研失信的撤销论文比例高于发达国家;在12类撤销原因中,近3年以作者出版失信、编辑/出版商失信和伪造/篡改论文3类涉嫌科研失信原因为主,在分布上存在着显著差异。【结论】为有效应对日益多变且隐蔽的科研失信行为,应结合撤销原因的特征,针对不同学科、国家和撤销原因情景,制定专门的管理策略,优化监测方案,并建立有效的识别与处置机制,提升学术界应对科研失信的整体能力。

关键词: 科研诚信, 撤销论文, 科研失信行为, 科研不端行为, Retraction Watch, 科研管理

Abstract:

[Purposes] The retraction of academic papers serves as a self-correction mechanism in the scientific community. Therefore, it is necessary to analyze the characteristics of retracted paper, so to reveal the history, current situation, and regular patterns of retraction, thereby providing targeted recommendations to enhance research integrity management for various academic communities. [Methods] Using global retracted paper records in Retraction Watch (RW), 32868 journal papers retracted between 1980 and 2023 were selected. The retracted reasons were divided into two major categories, “misconduct suspected” and “not misconduct”, with a total of 12 subcategories. Statistics were conducted from five aspects, including the time lag, reasons, disciplines, journals, and countries/regions of retraction. And the performance differences of the 12 subcategories of retraction reasons were analyzed in detail. [Findings] The number of retracted papers worldwide has consistently increased, with a notable rise in cases attributed to ‘misconduct suspected’. The average retraction lag in the past decade was approximately three years, with a shortening trend over the last five years. Basic life sciences (BLS) and health sciences (HSC) dominate as the most affected disciplines-‘hardest hit areas’ for retracted papers, while retracted papers, in business/technical (B/T) fields show increasing trends. The number of retracted papers from international collaborations is slightly lower than that from non-international collaborations. While the proportion of suspected misconduct papers is higher in developing countries than in developed ones. Among the 12 retracted reasons, the leading categories in the past three years are the authorial breach of trust, the editor/publisher’s breach of trust, and the forgery/tampering of papers, with significant differences observed in their distribution. [Conclusions] In order to effectively deal with the increasingly research misconducts, tailored management strategies should be formulated in the different disciplines, journals, and countries in combination with the characteristics of the retracted reasons. And we also need to optimize the method of monitoring, identifying and handling risky papers, to help the academic community to deal with dishonesty in scientific research.

Key words: Research integrity, Retracted paper, Research dishonest conduct, Research misconduct, Retraction Watch, Research management