用蒙特卡罗方法求取广义线性混合模型之最大似然估计:应用于求取乳癌死亡率之小区域估计（英文）

Maximum Likelihood Estimation in Generalized Linear Mixed Models Using Monte Carlo Methods: Application to Small-Area Estimation of Breast Cancer Mortality

本文使用蒙特卡罗方法,求得广义线性混合模型之最大似然估计,并提供用来评估统计参数之收敛和精确度之实用方法．仿真研究显示无偏之固定效应参数估计,而方差分量估计之误差则相近于前人结果．应用举例为使用泊松分布求取乳癌死亡率之小区域估计．

This paper provides a Monte Carlo approach for achieving maximum likelihood in generalized linear mixed models. Practical approaches for accessing convergence and precision of parameters are also discussed. Simulatioa study shows unbiased fixed effects parameter estimation with covariance components estimatioll comparable to previous study. Application for a small-area estimation of breast cancer mortality using Poisson distribution is illustrated.