雖然Greenwood估計值(GE)低估Kaplan-Meier估計值(KME)的變方, 但仍廣為使用。為了修正GE變方估計的偏差.Zhao(1996)提出另一估計 值,稱之為均勻估計值(homogenetic estimate,簡稱為HE)。本文中,我們 指出HE實際上估計還原估計值(reduced sample estimator,簡稱為RE,參 閱David(1993))的變方,且會嚴重高估KME的變方。我們推導HE和GE 之間的關係式。此外,提出在使用HE時應注意的一些問題。 關鍵詞：Greenwood估計值,均勻估計值,還原估計值。 The Greenwood estimate (GE) is commonly employed for estimating the variance of the Kaplan-Meier estimate (KME) even though it underestimates the variance. To correct the bias of GE , Zhao (1996) proposed an alternative, called the homogenetic estimate (HE). In this note, we point out that HE actually estimates the variance of the reduced sample estimate (RE) and can seriously overestimate that of KME. We also derive the explict relation between HE and GE and give some notes on the use of HE. KEYWORDS: Greenwood estimate;homogenetic estimate;reduced sample estimate.
