医学统计方法常用t检验处理两个样本均数间的差别显著性,但t检验要求资料为正态或接近正态分布才能应用,而医学实际工作中常遇到很多非正态分布的实例,就先要用各种代换数据处理,使资料变成正态分布,然后用代换数据作t检验,但往往有些资料找不到适当的代换数据,这样计算方法就变成复杂麻烦。近来从国外传入“非参数统计方法”,对正态与非正态资料都可用,但这些“非参数统计方法”也有缺点,对正态分布资料比不上t检验精确灵敏,希望大家研究出一种更好的非参数统计方法,对正态分布的资料,能够做到与 Medical statistical methods commonly used t test to deal with the significance of the difference between the two sample mean, but the t test requires the data to be normal or near normal distribution in order to apply, and medical practice often encounter many non-normal distribution of examples First use a variety of substitution data processing, the data becomes a normal distribution, and then use the substitution data for t test, but often some information can not find the appropriate substitution data, so that the calculation method becomes complicated trouble. Recently, “nonparametric statistical methods” have been introduced from abroad, and both normal and non-normal data are available. However, these “non-parametric statistical methods” also have disadvantages. They are not as accurate and accurate as the t-test for normal distribution data A better nonparametric statistical method can be achieved for normally distributed data