多性状综合评定的秩和差测验方法

A Rank-Sum-Difference Testing Method for Multi-Trait Comprehensive Ranking

【目的】多性状综合评价的秩和测验方法,虽可对多性状评价对象进行综合排序,并测验各评价对象与其平均水平的差异显著性,但无法实现各评价对象两两之间的差异显著性测验。为此,本文研究多性状综合评价多重比较的统计测验方法。【方法】在"H0:各评价对象在各性状上的秩次随机分布"假设下,以秩和理论分布为基础,利用组合数学方法。【结果】进一步导出了多性状秩和差的理论分布,并据之给出了多性状综合评价秩和差测验的显著性临界值。【结论】通过秩和差测验,确定评价对象两两之间的差异显著性,从而实现评价对象间的多重比较。该方法操作简单,便于实际应用。

【Objective】 The former paper had proposed the rank-sum testing method for multi-trait comprehensive evaluation,which not only can make the multi-trait comprehensive ranking,but also test the significance of difference between evaluation objects and the average level. In this paper,a rank-sum-difference testing method for multi-trait comprehensive evaluation was studied to implement the multiple comparisons among evaluation objects. 【 Method 】 The theoretical distribution of rank-sum-difference,based on the theoretical distribution of rank-sum,was firstly deduced by using combinatorial mathematics. 【Result】 Furthermore,it was used to obtain the critical values for the rank-sum-difference testing in multi-trait comprehensive evaluation. 【Conclusion】 The proposed rank-sum-difference testing method is so simple and convenient that it can be easily used to realize multiple comparisons among various evaluation objects.