• •

### 改进样本代表性的多目标追加平衡设计

• 出版日期:2018-12-25 发布日期:2018-12-28

### A Balanced Design with Added-up Multi-purpose Sample to Improve Sample Representatives

Gong Hongyu & Chen Ya

• Online:2018-12-25 Published:2018-12-28

Abstract: This paper mainly deals with two issues, i.e., improvement of sample representatives and multi-purpose sample survey. This paper proposes a new multi-purpose sampling methodology to improve sample representatives, an integration of added-up sample size and adjustment of sample composition, i.e., to create a new balanced sample from the original sample integrated with the added-up sample. In comparison with the original sample, the new integrated sample has reduced the composition bias between the original sample and the population. The balanced sample refers to Horvitz Thompson Estimator based on total auxiliary variables equal to the real value of the population total. Furthermore, a set of balanced samples can offer good representatives for various purposive parameters by selecting auxiliary variables related to the multi-purpose parameters and perform well for the multi-purpose sample survey. Based on the balanced sample with added-up using census data by county from the Sixth Population Census in 2010, a post facto assessment using multi-purpose parameters shows that the balanced design with add-up sample can improve sample composition in an effective way, make the sample and the population compositions close enough and reduces the errors in the purposive estimation. At the same time, the balanced sample design is justified for the multi-purpose survey with enhanced sample efficiency.