关键词: 多主题抽样调查, 最佳线性模型无偏估计, 一般回归估计, 相关性

Abstract: In the statistical business of sample surveys, there usually cover two or more target variables(indicators) in one survey program which is a typical multi-purpose survey. When estimating the target variables, we intend to estimate these two or more target variables simultaneously. Usually there are correlations among those surveyed variables, and it could be utilized to improve the estimation for target variables of interest by taking advantage of correlations among surveyed variables. This paper investigates the best linear model-based unbiased estimator and the general regression estimator under the multivariate multiple linear regression model. Compared with the most frequently used model of single response multivariate linear regression, the above two estimators for population total could effectively improve estimation precision. It is also elaborated by the results from the numerical simulation and a real case.

Key words: Multi-Purpose Sample Surveys, Best Linear Model-Based Unbiased Estimator , General Regression Estimator, Correlation