论文部分内容阅读
本文提出了配对计数资料的tc检验法,以克服配对四格表卡方检验中未考虑样本含量之缺欠。文中介绍了统计量tc的计算及推导;又对不同样本含量的配对四格表分别用卡方检验及tc检验法进行了比较,其差别说明了样本含量对检验结果有所影响。当检验显著性水平为0.05时,样本含量的影响不明显。可以应用配对四格表检验法或tc检验法;但当检验显著性水平为0.01时,则影响比较明显,此时以应用tc检验法更准确。tc检验法还可用于配对K×K表,本文举例说明了用法。
This paper presents the tc test of paired-count data to overcome the lack of sample content in the paired squares chi-square test. In this paper, the calculation and derivation of statistic tc are introduced. The paired squares with different sample contents are compared by chi-square test and tc test respectively. The difference indicates that the sample content has an effect on the test result. When the level of significance of the test was 0.05, the influence of the sample content was not significant. Paired four-squared table test or tc test can be applied; however, when the significance level of the test is 0.01, the effect is more obvious. At this time, the application of tc test is more accurate. The tc test can also be used to pair K×K tables. This article illustrates the usage.