摘要：

【目的】 分析被撤销医学论文中数据学术不端类型及其行为特征,提出相应建议,为识别和防范此类学术不端行为提供借鉴。【方法】 选择Retraction Watch数据库中1902篇被撤销医学论文作为数据源,分析论文被撤销的主要原因,总结医学论文数据学术不端的常见类型和行为特征,从作者、作者单位、期刊出版单位、制度建设等方面提出相应的防范措施。【结果】 2005年以后因数据问题被撤销的医学论文数量呈较快增长趋势,并在国家/地区分布上呈集中趋势(54.47%分布于美国、日本、中国);伪造数据、存疑数据、错误数据、不可靠数据、重复数据、剽窃数据等是被撤稿的主要原因,主要可归纳为数据伪造、数据剽窃、数据篡改3种类型,其呈现方式多样且行为特征各异,准确把握其基本特征,有助于科学辨识、防范数据学术不端行为。【结论】 作者主动把好“源头关”,自觉接受诚信监督;作者单位履行科研诚信建设主体责任;期刊出版单位提高数据存疑意识,强化数据和统计学审查,建立多技术协同防范机制;多部门主动构建失信联合惩戒机制,以有效遏制数据学术不端行为。

关键词: 学术不端, 数据伪造, 数据剽窃, 数据篡改, 行为特征, 被撤销医学论文

Abstract:

[Purposes] This paper aims to analyze the types and behavior characteristics of data academic misconduct in retracted medical articles, and put forward corresponding suggestions to identify and prevent this academic misconduct. [Methods] A total of 1902 retracted medical articles in Retraction Watch database were selected as data sources, to analyze the main reasons for the retracted articles, summarize the common types and behavioral characteristics of data academic misconduct in retracted medical articles, and put forward corresponding precautionary measures from the aspects of author, author affiliation, periodical publisher, and institutional improvement. [Findings] Since 2005, the number of retracted medical articles due to the data academic misconduct has shown a rapid growth trend, and there is a trend of concentration in the distribution of countries/regions (54.47% in the United States, Japan, and China). The main reasons for the retracted articles are fabrication of data, unreliable data, concerns about data, error in data, duplication of data, and plagiarism of data. It can be divided into three types: fabrication of data, plagiarism of data, and manipulation of data, which has various expression and different behavioral characteristics. Accurately grasping its basic characteristics is helpful to scientifically identify and prevent data academic misconduct. [Conclusions] It is expected to effectively reduce the data academic misconduct while the authors consciously control the source and accept the honest supervision; the author affiliation fulfills the main responsibility of scientific research integrity construction; the periodical publisher should improve the awareness of data doubt, strengthen the examination of data and statistics, and establish the prevention mechanism by multi-technology; and multiple departments actively build a joint punishment mechanism for dishonesty.

Key words: Academic misconduct, Fabrication of data, Plagiarism of data, Manipulation of data, Behavioral characteristics, Retracted medical article