住户调查是我国社会经济统计调查体系的重要组成部分,样本代表性直接决定统计数据质量。多阶段抽样中初级单元的方差对估计的影响是主要的,因此本文结合2010年全国第六次人口普查分县数据,采用平衡抽样设计获取初级单元的代表性样本-平衡样本。对代表性样本的事后评估结果表明,样本结构与总体结构吻合,目标估计的误差很小,说明了本文平衡设计的有效性。
Household surveys are an important part of the socio-economic survey system. Representative samples directly determine the quality of statistical data. The variance of primary unit has the main influence on the estimate in the multi-stage sampling. Basing on the data of the sixth national census in 2010, it uses the balanced sampling for getting the balanced samples which are the representative samples of primary units. By assessing the representative samples, it shows that the sample structure is consistent with the population structure, and sampling error of target estimator is small, which indicates that the balanced design is effective.