基于回归组合技术的连续性抽样估计方法研究

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。

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

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.