一种基于多变量空间非概率抽样方法的设计  

A Design of Non-Probability Sampling Method Based on Multivariable Space

作　　者：张维群[1] 余欣媛 赵鲲鹏[2] 

机构地区：[1]西安财经学院统计学院,西安710100 [2]西安交通大学经济与金融学院,西安710061

出　　处：《统计与决策》2017年第20期76-78,共3页Statistics & Decision

基　　金：国家哲学社会科学基金资助项目(13TJ006);全国统计科学研究重点项目(2012LZ020);陕西省教育厅重点实验室科学研究项目(13JZ023)

摘　　要：文章基于样本均值无偏估计和有效估计的前提下,以行政区域单位划分为抽样框,采用多个辅助指标控制多目标抽样估计的误差,设计了一种多指标空间非概率抽样样本选取方法。并利用陕西省2014年城镇人均可支配收入以及人口增长率两个辅助指标对陕西省107个区县进行了抽样应用,结果显示样本的抽取涵盖了各个不同水平层次的区县,抽样效果良好。With sample mean unhiased estimation and effective estimation as a premise, this paper selects administrative re- gion as the sampling box, and uses muhiple auxiliary, indexes to control the error of multi-objective sampling estimation. Mean- while the paper designs a sample selection method of multi-index space non-probahility sampling, and also makes use of two aux- iliary indexes, namely the urban per capita disposable income and growth rate of population of Shaanxi Province in 2014, to carry. out a sample applieatiun to its 107 counties. The study results show that the sample extraction eovers the districts and counties at different development levels, and that the sampling effect is benign.

关 键 词：多变量 空间样本 抽样设计 抽样方法 

分 类 号：C81[社会学—统计学]