基于最终类集的复杂样本设计研究  

The Research of Complex Sample Design Based on the Ultimate Clusters

作　　者：罗薇[1,2] 刘建平[3] 

机构地区：[1]广东工业大学管理学院,广东广州510520 [2]广东工业大学大数据战略研究院,广东广州510006 [3]暨南大学经济学院,广东广州510632

出　　处：《数理统计与管理》2017年第3期469-480,共12页Journal of Applied Statistics and Management

基　　金：国家社科基金重点项目(12ATJ002);广东省哲学社会科学"十二五"规划2015年度学科共建项目(GD15XGL09)

摘　　要：本文依据最终类集样本可以近似均等选择概率下各种多阶段抽样设计样本的理论,构建出多阶段抽样下复杂样本设计的理论分析框架,解析出影响复杂样本估计量设计效应的两个基本因素:最终类集同质系数和最终类集规模,分析了它们对复杂样本估计量设计效应的影响机理及相互作用机制,并应用于我国一体化住户调查的主样本设计。An ultimate clusters sample can serve as an approximation for a complex equal probability of selection method multi-stage sample design.According to the approximation, it provides a theoretical analysis framework for complex sample under multi-stage sample design. Within this framework, the complex sample estimator design effects represent a number of components, including the intra-cluster correlation and cluster size of ultimate clusters. The mechanism of the effects of these components and their interactions are measured for the design effects of complex sample estimator, and used for the master sample design of the integrate system of household surveys in China.

关 键 词：最终类集 复杂样本 设计效应 

分 类 号：O212[理学—概率论与数理统计]