基于逻辑自洽的科技论文中样本量夸大行为的统计学甄别  

Statistical identification of sample size inflation in scientific papers based on logical consistency

作　　者：周涌 胡莉 周皎石[3] ZHOU Yong;HU Li;ZHOU Jiaoshi(School of Mathematics and Statistics,Hubei University of Science and Technology,437100,Xianning,Hubei,China;不详)

机构地区：[1]湖北科技学院数学与统计学院,湖北咸宁437100 [2]湖北科技学院学报编辑部,湖北咸宁437100 [3]湖北科技学院生物医学工程与医学影像学院,湖北咸宁437100

出　　处：《编辑学报》2025年第1期70-74,共5页Acta Editologica

摘　　要：大样本量数据兼具重要学术价值和不易获取2种特性,使得一些科技论文中出现了样本量夸大这种学术不端行为,且具有高度隐蔽性,难以被侦测。科技论文中常常需要展示数据多方面的特征,这些特征相互之间会呈现关联性和协调性。如果论文中存在样本量夸大行为,各特征之间的关联性和协调性将会被打破而无法形成逻辑自洽。本研究旨在通过Monte-Carlo模拟等技术手段,侦测论文中是否存在无法形成数理逻辑自洽的统计特征,借以判断文中是否存在样本量夸大行为。以几篇待审稿件和已发表论文为例进行分析,结果表明这些技术手段不仅能甄别出样本量夸大行为,还能测算出夸大幅度。故建议对于含有数据分析模块的论文,需要有精通数理逻辑的专家作为联合审稿人。Large-size samples possess both significant academic value and difficulty in acquisition,leading some scientific papers to exhibit sample size inflation,a form of academic misconduct that is highly concealed and difficult to detect.Since scientific papers often need to showcase various data characteristics,these features exhibit interrelationships and coordination.If there is sample size inflation,the relationships and coordination among features will be disrupted,resulting in a lack of logical consistency.This study aims to detect whether a paper contains statistically inconsistent features using techniques such as Monte-Carlo simulations,thereby identifying sample size inflation.Analysis of several manuscript drafts and published papers indicates that these techniques can not only detect sample size inflation but also estimate the extent of the inflation.It is recommended that documents with data analysis modules should have experts proficient in mathematical logic as co-reviewers.

关 键 词：样本量夸大 甄别 逻辑自洽 MONTE-CARLO模拟 

分 类 号：G232[文化科学]