相关四格表数据检验需要注意的几个问题  

Some Problems Needing Attention in Data Inspection of Relevant Fourfold Table

作　　者：陆运清[1] Lu Yunqing(College of Education,Hebei Normal University,Shijiazhuang 050024,China)

机构地区：[1]河北师范大学教育学院,石家庄050024

出　　处：《统计与决策》2022年第7期31-34,共4页Statistics & Decision

基　　金：河北省社会科学基金资助项目(HB20JY006)。

摘　　要：相关四格表在实际研究中应用广泛,但存在的几个典型问题影响了分析结果的正确性和准确性。这些问题主要包括:(1)统计著作及软件的有关分析方法存在丢失部分样本信息的问题,应用中需要关注有关方法的研究进展,采用更为完善的方法;(2)相关四格表和独立四格表相混淆的问题。相关四格表和独立四格表的问题实质、假设、公式及适用条件等均不相同,在实际应用中需要选用正确的方法;(3)将相关四格表转化为独立四格表检验两次测试结果的差异。由于相关四格表转化为独立四格表的过程中未考虑两次测试的对象为同一组被试,即未考虑两次测试的相关,导致卡方值翻倍。文章对以上问题进行了甄别分析。Relevant fourfold table is widely used in practical research, but several typical problems affect the correctness and accuracy of analysis results. These problems mainly include the following:(1) Some sample information is lost in the analysis methods of statistical works and software, so it is necessary to pay attention to the research progress of relevant methods and adopt more perfect methods in the application.(2) Confusion exists between relevant and independent fourfold table. The essence of the problem, hypothesis, formula and applicable conditions of the relevant and independent fourfold tables are different, so it is necessary to choose the correct method in practical application.(3) The relevant fourfold table is converted into an independent fourfold table to test the difference between the two test results. The relevant fourfold table converted into an independent fourfold table for testing, there exists difference between the two results. Since the same group of subjects are not considered in the process of transforming the related fourfold table into an independent fourfold table, that is, the correlation between the two tests is not considered, resulting in the doubling of the Chi-square value. The paper discriminates and analyzes the above problems.

关 键 词：相关四格表 独立四格表 数据检验 统计量膨胀 

分 类 号：O212.7[理学—概率论与数理统计]