摘要
小域估计(SAE)是小样本抽样调查中的重要议题,该问题下使用经典统计学方法的估计偏差较大。文章基于经验贝叶斯理论的基本思想构建分层贝叶斯模型,以我国肺结核疾病死亡率为例提出了一种小样本条件下有关健康比率数据的小域估计方法,应用MCMC算法对模型中的参数和超参数进行了后验推断。结果表明,与传统方法相比,分层贝叶斯模型同时考虑了不同地区间的同质性与异质性,并可根据样本数量大小对小域估计量进行合理调整,其估计结果更具稳健性。
Small area estimation(SAE)is an important issue in the small sample survey,in which using classical statistical methods has a large estimation deviation.This paper is based on the basic idea of empirical Bayesian theory to construct a hierarchical Bayesian model,taking the mortality rate of tuberculosis in China as an example to present a SAE method for health ratio data with small sample size,using MCMC algorithm to carry on posterior inference of parameters and super-parameters in the model.The results show that compared with the traditional methods,hierarchical Bayesian model considers the homogeneity and heterogeneity of different regions,and can adjust the estimators of small area according to the size of samples,and its estimation results are more robust.
作者
张圆
周兰兰
Zhang Yuan;Zhou Lanlan(School of Statistics,Tianjin University of Finance and Economics,Tianjin 300222,China;Hebei Branch of PICC Property and Casualty Company Limited,Shijiazhuang 050000,China)
出处
《统计与决策》
CSSCI
北大核心
2019年第19期15-19,共5页
Statistics & Decision
基金
国家社会科学基金资助项目(16BTJ001)
关键词
小域估计
经验贝叶斯
分层模型
疾病死亡率
small area estimation
empirical Bayes
hierarchical model
disease mortality
