期刊影响因子的加权信度估计及统计分析  

Credibility Weighted Estimation and Statistical Analysis of Journal Impact Factors

作　　者：温利民[1] 黄根帆 章溢 WEN Limin;HUANG Genfan;ZHANG Yi(School of Mathematics and Statistics,Jiangxi Normal University,Nanchang 330022;School of Finance,Jiangxi Normal University,Nanchang 330022)

机构地区：[1]江西师范大学数学与统计学院,南昌330022 [2]江西师范大学财政金融学院,南昌330022

出　　处：《系统科学与数学》2023年第7期1904-1920,共17页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(72263019);江西省自然科学基金(20224BCD41001);江西省研究生创新基金(YC2022-s285)资助课题。

摘　　要：期刊影响因子是度量期刊质量的重要标准.然而,影响因子估计期刊质量时仅仅利用了引用次数的样本信息,而忽略了影响期刊质量的先验信息.文章建立了期刊质量参数的贝叶斯模型,将期刊质量参数的估计限定在加权引用次数的线性组合中,得到了期刊质量参数的最优加权线性估计.结论表明,最优估计能表达为加权影响因子与聚合估计的加权平均,其中权重满足“信度估计”的性质.信度估计不仅具有很好的统计性质,而且能对传统的期刊影响因子进行了校正,更加客观地度量期刊的质量水平.The journal impact factor is an important standard to measure journal quality.However,the impact factor only uses the sample information of the number of citations to estimate the quality of journals,and ignores the prior information that affects the quality of journals.In this paper,the Bayesian model of journal quality parameters is established,and the estimation of journal quality parameters is limited to the linear combination of the weighted number of citations,and the optimal linear Bayesian estimation of journal quality parameters is obtained.The conclusion shows that the optimal estimation can be expressed as the weighted average of the weighted impact factor and the aggregation estimation,in which the weight satisfies the property of "credibility".The credibility estimation not only has a good statistical property,but also can correct the traditional journal impact factors,and more objectively measure the quality level of journals.

关 键 词：期刊影响因子 质量参数 线性贝叶斯估计 信度估计 相合性 渐近正态性 

分 类 号：O212.1[理学—概率论与数理统计] G237.5[理学—数学]