一般统计检查法,常强调样本数据须符合某种假定,如相比较的样本是来自正态、或接近正态分布的总体,又它们的方差(或称变异数)不能相差悬殊(必要时,要作方差一致性检验)等。决定正态分布特征的统计指标是总体平均数与标准差,t 检验等须要利用这些参数。因此,这些统计方法,可称参数检验法。但在实际工作中,有时不知道数据是什么分布;或数据不符合上述假定,如从明显的非正态的总体中抽得的小样本,或方差相差悬殊、同质性较差的样本资料,通常的参数检验法就不一定适用。这时,须要一种不依赖于某种特定分布的检验方法,这就是非 The general statistical inspection method often emphasizes that the sample data must meet certain assumptions. For example, the compared samples are from the normal, or close to normal distribution of the population, and their variance (or variance) can not be significantly different (if necessary, To do variance consistency test) and so on. The statistical indicators that determine the characteristics of the normal distribution are the population mean and standard deviation, and the t-test, etc., must use these parameters. Therefore, these statistical methods can be called parametric test methods. However, in actual work, sometimes it is not known what the distribution of data is; or the data do not meet the above assumptions, such as small samples drawn from obvious non-normal populations, or sample data with poor variance and poor homogeneity. The usual parametric test may not apply. At this time, a test method that does not depend on a specific distribution is required.