。该法根据所研

究的问题构造检验统计量，并利用手头样本，按排列组合的原理导出检验统计量的理论抽样分布；

若难以导出确切的理论分布，则采用抽样模拟的方法估计其近似分布，然后求出从该分布中获得

手头样本及更极端样本的概率(P值)，并界定此概率值，作出推论。-It is sample．based distribution-free method，which makes use of“permutation or combination”to get the theo．

retical sampling distribution of the constmcted“statistics”，or get

the approximate distribution by simulation under

the nuU hypothesis，gets the pmbability of the sample-at—hand，and dmws a conclusion after comparing the probability

w“h the size of the test．It especiaUy supports

small samples without distributional assumptions(e．g．normdb2y)and

complex des培ns for which traditional methods are hard to solVe．