复杂抽样的Bootstrap方差估计方法及应用

Bootstrap Variance Estimation Method and Application for Complex Sampling

构造总体参数估计量的方差估计是抽样估计重要内容.但当抽样方案复杂或估计量形式复杂时,传统抽样调查的方差估计方法很难适用.本文探究适用于中国人口抽样调查的Bootstrap方差估计方法.为此,首先采用文献解读法,解析复杂抽样Bootstrap方差估计方法的基本理论;接着,提出适用于中国人口抽样调查的重权数Bootstrap方差估计方法与类总体Bootstrap方差估计方法,并对它们的统计性质进行仿真模拟试验;最后,利用天津市2015年全国1%人口抽样调查数据,演示两种Bootstrap方差估计方法的应用,验证其实践有效性.本文有助于提升中国在政府统计抽样估计领域的基础理论研究水平,并为中国人口抽样调查以及其他抽样调查的方案设计和估计提供借鉴.

Constructing variance estimator of population parameter estimator is an important part of sampling estimation.However,when the sampling scheme is complex or the representation of estimators is complex,the traditional variance estimation method is difficult to apply.This paper researches the bootstrap variance estimation method applicable to the Chinese population sampling survey.We firs tuse the literature interpret method to analyze the theory foundation of bootstrap variance estimation method for complex sampling.Secondly,the rescaled bootstrap variance estimation method and the pseudo-population bootstrap variance estimation method are revised to suitable for Chinese population sampling survey.A Monte Carlo simulation study was conducted to examine their statistical properties.Finally,using the 1%population sampling survey data of Tianjin in 2015,two bootstrap variance estimation methods is demonstrated to verify their effectiveness in application.This study can improve the theoretical base research level of the Chinese government's statistical sampling estimation,and provide reference for the design and estimation of Chinese population sampling survey and other sampling surveys.