“两类错误”谈论有益

PROFITABLE TO SAY “TWO TYPES OF ERRORS”

本文是作者９３年发表一文［２］的补充，也是对本刊９８年第二期一文［１］的商榷。文中重述了两种决策的不同可靠性，即在否定虚假设Ｈ０时可以控制出错率到指定要求，使结论可靠；而接受Ｈ０则不易做到这点。对被检验假设的否定或接受均通过否定Ｈ０来获得是最好的。叙述了设置Ｈ０和备择假设Ｈ１的考虑。指出当前应用中的主要问题是对Ｈ０的不适当接受，即对第二类错误的忽视。重温一些基本概念等。

This is a supplementary note to the author's previous paper and is also in response to a discussion on statistical test of hypothesis. The paper includes five parts entitled as Re-emphasis of two types of decisions resulting in different reliabilities, More comments on establishing H0 and H1, More comments on type II error, Reviewing some basic concepts, and Conclusion remarks.