在藥物的第二、三階段的臨床試驗時，藥物的安全性和有效性是測驗的主要指標。Hsu 和Berger在藥物各個劑量組的方差σ²都相等這一假設(shè)下，提出一種識別最小有效劑量的逐步置信區(qū)間方法。而方差齊性假設(shè)條件通常是不合理的，本文主要利用 Stein 兩階段抽樣方法，給出方差無任何限制條件下識別最小有效劑量（MED）的逐步置信區(qū)間方法。

It is very important to be guaranteed that drug’s safety and efficiency to the mankind in Phase II and III clinical trials. Recently Hsu and Berger (1999) proposed a stepwise confidence interval method in dose-finding of minimum effective dose (MED) for dose-response study under equal variances. This is often in doubt. A new way to find MED was proposed by employing the Stein’s two-stage sampling method, the stepwise confidence procedure was proposed without any condition imposed on the variance. And the error rate is properly controlled.