两两比较的Bonferroni法

The Bonferroni Method for Multiple Comparisons

几组资料的平均水平被初步判定差异有统计学意义后,往往需进一步做两两比较。两两比较时,最常见的错误是将总的检验水准(如琢=0.05)直接用于每一次的比较中。Bonferroni法通过修正每次比较的检验水准,控制Ⅰ类错误总的发生概率,是两两比较中最常用的方法之一。本文简要阐述了Bonferroni法的基本原理,并举例说明Bonferroni法的几种应用情况,旨在指导读者运用Bonferroni法正确地进行多组间的两两比较。

When the differences of average levels among several groups are statistically significant, the multiple comparisons should be carried out. Using α=0.05 directly as significance level in comparison between each pair may result in too high type I error probability. Bonferroni method is one of the most common methods for multiple comparisons, which could reduce the accumulated type 1 error probability by adjusting α level. In this report, the fundamental principle of Bonferroni method is explained and examples are presented in some specific situations. This article is supposed to guide researchers to utilize Bonferroni method correctly in multiple comparisons.