多变量分层抽样技术的改进与应用

The Improvement of Multi-variable Stratified Sampling and Its Application

多目标抽样调查对于提高样本利用效率、节约调查费用、全方位反映总体都有极为重要的意义。通过引进Ward系统聚类法改进多目标分层抽样方法;选择合适的调查指标作为辅助变量构造分层比率估计推断差异较大的指标,改进多目标分层估计量。最后进行实际应用与比较。总体上该改进的分层设计方案不仅在精度上、成本上有较大的优良性,而且在可操作性方面更显优势。

Multi-object sample survey is very important for the improvement of sample utilization efficiency, cost saving, and the omni-directional reflection of population as well. Both sampling methods and estimation of multivariable stratified sampling by introducing is improved into Ward＇s hierarchical clustering method and stratified ratio estimates with suitable auxiliary variables among the multi-object. In addition, the comparison with practical appli- cation shows its superiority on precision, cost and easy operation in multi-object sample survey.