多重比较及其在销售数据分析上的应用

Multiple Comparisons and its Application in Sales Data Analysis

本文对LSD方法、Bonferrioni方法、SNK方法、Tukey方法和Duncan方法等五种多重比较方法的检验效果进行了比较研究,并以Only服装销售数据为例,用SPSS软件中的多重比较方法对该数据进行了分析和比较。比较结果表明,LSD方法比Bonferroni方法有更高的灵敏度,却不如Bonferroni方法能更好地控制FWER,而划分相似子集时Duncan方法比SNK方法和Tukey方法更灵敏。

This paper makes a comparative study on the test effects of LSD method, Bonferroni method, SNK method, Tukey method and Duncan method. Taking Only garment marketing datas as example, the dates have been analyzed and compared by multiple comparisons method of SPSS software. The results shows that LSD method has higher sensitivity than Bonferroni method, but it can＇t control FWER better. Furthermore, Duncan method is more sensitive to similar subsets partition than SNK method and Tukey method.