综合评价结果的区分度与稳定性研究  

Research on Discriminability and Stability of Comprehensive Evaluation Results

作　　者：李兴奇[1] 高晓红[2] Li Xingqi;Gao Xiaohong(School of Management and Economics,Chuxiong Normal University,Chuxiong Yunnan 675000,China;School of Mathematics and Computer Science,Chuxiong Normal University,Chuxiong Yunnan 675000,China)

机构地区：[1]楚雄师范学院管理与经济学院,云南楚雄675000 [2]楚雄师范学院数学与计算机科学学院,云南楚雄675000

出　　处：《统计与决策》2022年第16期16-21,共6页Statistics & Decision

基　　金：国家自然科学基金资助项目(11261001);云南省应用基础研究计划青年项目(2017FD152)。

摘　　要：综合评价是一项系统而又复杂的工作,其目的是获取能反映被评价对象全貌的综合评价值来区分不同个体的优劣。好的综合评价结果应该具备较高的区分度和稳定性,因此如何科学度量评价值的区分度和稳定性成为综合评价结果可靠与否的关键。文章构建一种新的区分度度量方法,提升了区分度的准确性和适用性;分析了导致综合评价结果波动的诸多因素,引进综合评价结果稳定性的度量指标,并证明了该指标的科学性与合理性;通过大量的数值模拟实验来计算不同综合评价方法的区分度和稳定性,研究区分度和稳定性的影响因素。结果表明,不同分布类型数据的取值规律不同,不同无量纲化方法处理数据的效用不同,不同综合评价方法确定权重的机理不同,任何一种因素的改变都会对综合评价结果的区分度和稳定性产生影响,所以进行综合评价前需要充分考虑数据的分布类型进而选择恰当的无量纲化方法和综合评价方法,以提高结果的区分度和稳定性。Comprehensive evaluation is a systematic and complicated task, and its purpose is to obtain the comprehensive evaluation value that can reflect the whole picture of the evaluated object to distinguish the advantages and disadvantages of different individuals. A good comprehensive evaluation result should have a high discriminability and stability, so how to scientifically measure discriminability and stability of evaluation value becomes the key to the reliability of comprehensive evaluation results.This paper constructs a new method to measure the discriminability, which improves the accuracy and applicability of discrimination. Then the paper analyzes the factors that lead to the fluctuation of the comprehensive evaluation results, introduces the measure index of comprehensive evaluation result stability, and also proves the scientificity and rationality of the index. Finally, the paper calculates the discriminative degree and stability of different comprehensive evaluation methods through a large number of numerical simulation experiments, and makes a study on the influencing factors of discriminative degree and stability as well. The results go as below: Different distribution types of data have different value rules;different dimensionless methods have different utility in processing data, and different comprehensive evaluation methods have different mechanism in determining weight. Therefore, before comprehensive evaluation is made, it is necessary to fully consider the distribution type of data and select appropriate dimensionless methods and comprehensive evaluation methods to improve the discriminability and stability of the results.

关 键 词：综合评价 区分度 稳定性 无量纲化 数值模拟 

分 类 号：C812[社会学—统计学] O213.9[理学—概率论与数理统计]