A surprising MLE for interval-censored binomial data

被引：3

作者：

Frey, Jesse ^{[1
]}

Marrero, Osvaldo ^{[1
]}

机构：

[1] Villanova Univ, Dept Math Sci, Villanova, PA 19085 USA

来源：

AMERICAN STATISTICIAN | 2008年 / 62卷 / 02期

关键词：

binomial distribution; exponential families; interval-censored data; log odds;

D O I：

10.1198/000313008X305034

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

We derive the maximum likelihood estimator of the binomial success probability when the number of trials is known, but the number of successes is only known to fall in some interval. This estimator has a surprisingly simple form. Specifically, the maximum likelihood estimate of the log odds of success is the mean of the maximum likelihood estimates of the log odds of success that are obtained by letting the number of successes take on each integer value in the interval. An application is suggested.

引用

页码：135 / 137

页数：3