On Three Error Measurements In Multiple Hypothesis Testing

Hypothesis testing is one of the most important aspects in statistic infer-enc.The Single hypothesis testing theory has been improved . The basic idea of the single hypothesis testing is to control the type I error and then to find a kind of rule so that the type II error is as small as possibe and then the testing is as powerful as possible. In some cases, wo can gain the most powerful tests, uniformly most powerful tests and most uniformly powfer unbiased tests etc for single testing. However, for the multiple hypothesis tcsting,no such result can be gained. With the rapid development of nature science, plenty of data can be gained. How to analysis these data become the challenge for statists and promote the devel-opmentof multiple hypothesis testing because the multiple hypothesis testing is the basic theroy for these data. For example, as the analysis of DNA microarray ,we want to know which genes are different in the expresion levels of thousands of genes which involves the multiple hypothesis testing problem. The essay main review the thoery and methods of multiple hypothesis testing and summarize the main ideas and methods acording to three kindsof type I error measure-FWER, FDR 和 pFDR, expatiate their difference, relation , weakness and the development in the future.