摘要
[研究目的]在指标体系学术评价中,传统意义上比较重视指标的整体权重问题,而较少关注评价对象个体权重问题,并且普遍存在专家权重(设计权重)与实际权重的偏离问题,从而使得专家权重失效,这就是伪权重问题。[研究方法]以JCR2019经济学期刊为例,在分析整体权重与个体权重的基础上,提出采用动态最大均值标准化方法与sigmoid函数标准化方法来消除整体伪权重问题,并对个体权重问题进行了深入分析。[研究结论]对于整体线性评价而言存在设计权重与实际权重不相等的伪权重问题;sigmoid函数标准化与动态最大均值逼近标准化方法均可以有效解决伪权重问题;个体评价中广泛存在伪权重问题并且理论上不可能消除;解决整体评价的伪权重问题是讨论个体评价权重问题的基础;要辩证看待评价的权重导向作用。
[Research purpose]In the academic evaluation of the index system,the traditional sense pays more attention to the overall weight of the index,but less attention to the individual weight of the evaluation object,and there is a common deviation between the expert weight(design weight)and the actual weight,which makes the expert weight invalid.This is the pseudo weight problem.[Research method]Taking JCR2019 economic journal as an example,based on the analysis of the overall weight and individual weight,this paper proposes the dynamic maximum mean standardization method and sigmoid function standardization method to eliminate the overall pseudo weight problem,and makes an in-depth analysis of the individual weight problem.[Research conclusion]The research shows that:there is a pseudo weight problem that the design weight is not equal to the actual weight in the overall linear evaluation.Sigmoid function standardization and dynamic maximum mean approximation standardization methods can effectively solve the pseudo weight problem.There is a wide range of pseudo weight problems in individual evaluation,and it is impossible to eliminate them in theory.To solve the problem of pseudo weight of overall evaluation is the basis of discussing the problem of individual evaluation weight.We should dialectically treat the weight guiding role of evaluation.
作者
俞立平
潘伟波
Yu Liping;Pan Weibo(School of Statistics and Mathematics,Zhejiang Gongshang University,Hangzhou 310018;Collaborative Innovation Center of Statistical Data Engineering,Technology&Application,Zhejiang Gongshang University,Hangzhou 310018)
出处
《情报杂志》
CSSCI
北大核心
2022年第10期163-169,198,共8页
Journal of Intelligence
基金
浙江省重点建设高校优势特色学科(浙江工商大学统计学)资助(编号:2021A18)。
