摘要多阶段抽样技术广泛应用于流行病学现况调查中,但针对其所得资料的统计分析方法往往选择不当.文中介绍一种用于多阶段抽样调查资料的统计分析方法--加权估计法,以推广针对此类资料的恰当的分析方法.在介绍加权估计法基本原理的基础上通过两个二阶段分层整群抽样的实际调查资料实现这种算法.加权估计法可以校正由多阶段抽样导致的三种效应:群效应、分层效应、不等概率性,给出无偏点估计和比较客观的误差估计,并作出正确的统计推断.

abstractsMultistage sampling techniques are widely applied in the cross-sectional study of epidemiology, while methods based on independent assumption are still used to analyze such complex survey data. This paper aims to introduce the application of weighted estimation methods for the complex survey data. A brief overview of basic theory is described, and then a practical analysis is illustrated to apply to the weighted estimation algorithm in a stratified two-stage clustered sampling data. For multistage sampling survey data, weighted estimation method can be used to obtain unbiased point estimation and more reasonable variance estimation, and so make proper statistical inference by correcting the clustering, stratification and unequal probability effects.