关于样本均值的抽样分布能否作正态近似的探讨

一、引言当我们对总体均值进行统计推断时,常常需要假定样本均值服从或近似服从正态分布。我们知道,当样本来自于正态总体时,样本①均值服从正态分布;当样本来自于非正态总体时,根据中心极限定理对于足够大的样本容量n,样本均值将近似服从正态分布。对于非正态总体,问题的关键是样本容量n的“足够大”到底指多少?这很难一概而论。人们通常以30为界,将n≥30的样本称为大样本,并认为样本均值-x的抽样分布可作正态近似;而将n<30的样本称为小样本,认为此时不宜将-x的抽样分布作正态近似。许多统计应用者都是按这样的工作规则来做的,但在许多实际…

The Argument whether Sampling Distribution of Expectation of Samples Approximately Obey Normal Distribution

Based on the concepts of skewness and kurtosis, the paper points out a kind of method to judge whether the sampling distribution of sample mean can be normally approximated. Besides, the paper also provides the discussion of how to make judgement uinder the following conditions: 1. Finite population with outlier; 2. Unknown population distribution.