基于回归组合技术的连续性抽样估计方法研究  

A Study on Successive Sampling Estimation Based on Regression Composite Estimation Technique

作　　者：刘建平[1] 陈光慧[1] 

机构地区：[1]暨南大学经济学院,广东广州510632

出　　处：《统计与信息论坛》2008年第10期5-8,共4页Journal of Statistics and Information

基　　金：国家社会科学基金项目<连续性抽样调查方法及其应用研究>(08BTJ010);全国统计科学研究重点项目<建立连续性抽样调查体系的问题研究>(2007LZ038)

摘　　要：在使用样本轮换的连续性抽样调查中,不仅可以利用前期调查的研究变量的信息,还可使用现期调查的辅助变量信息来建立回归模型进行回归估计,进而构造回归组合估计量,并在此基础上确定最优样本轮换率和最优权重系数,使得回归组合估计量的方差最小,从而更大程度地提高连续性抽样调查的估计精度。This paper discusses how to improve the efficiency of sample rotation under successive sampling when some auxiliary information can be obtained. We can not only make use of the information of study variable acquired by the last survey, but also the information of auxiliary variable in current occasion. Then, regression models can be established through the correlation of study variable and auxiliary variable and regression estimators are obtained. Furthermore, regression composite estimator is constructed by these regression estimators. The optimal ratio of sample rotation and coefficient of weight can be confirmed which can minimize the variance of regression composite estimator. Therefore, the efficiency of successive sampling can be improved more through the method of regression composite estimation introduced by this paper.

关 键 词：连续性抽样 样本轮换 辅助变量 回归组合估计量 

分 类 号：C811[社会学—统计学]