摘要
对“三新”企业进行抽样调查是及时掌握和监测“三新”经济发展的重要手段。考虑到这一类调查总体单元变动比较迅速,抽样框信息变动大,无法及时覆盖总体的最新特征,依此抽样框得到的样本数据结构与总体的分布结构差异较大,样本的代表性较低,会对总体数量特征的有效估计产生影响。因此,基于调查总体单元的变动特征,把抽样框中的单元划分为保留单元和转移单元,在此基础上,依据样本单位分层结构的变动,设计了基于“三新”企业分层抽样单元权重动态调整的估计方法。首先,通过事后分层方法挖掘出不同层的单位特征,并预测抽样框各层容量;其次,依据层规模的变动预测对目标变量估计量的权重进行修正;最后,通过自我加权设计构造出总体动态变动后数量特征的复合估计量,并对其进行优良性讨论。在对“三新”企业的模拟数据进行多次重复抽样实验中,相比于固定抽样框下的传统方法,基于分层抽样单元权重动态调整的估计方法具有更高的抽样效率,构造的关于总体数量特征的估计量具有无偏性和有效性。
Sampling survey of"three new"enterprises is an important method to timely grasp and monitor the economic development of"three new".Considering the population unit of the survey changes rapidly,the sampling frame information changes greatly,and the latest characteristics of the population cannot be covered in time.The sample data structure obtained by this sampling frame is quite different from the distribution structure of the population,and the representativeness of the sample is relatively weak,which will affect the effective estimation of the population quantitative characteristics.Based on the changed characteristics of survey population units,the units in sampling frame are divided into reserved units and transfered units.On the basis,according to the change of stratified structure of sample units,an estimation method based on dynamic adjustment of stratified sampling unit weight of"three new"enterprises is designed.Firstly,the unit features of different layers are mined by post-stratification method,and the capacity of each layer of sampling frame is predicted.Secondly,the weights of the target variable estimators are modified according to the prediction of the change of the layer size.Finally,compound estimators of quantitative characteristics after population dynamic changes are constructed by self-weighted design,and their optimality is discussed.Compared with the traditional method under fixed sampling frame,the estimation method based on dynamic adjustment of stratified sampling unit weight has higher sampling efficiency,and the constructed estimators of population quantitative characteristics have unbiased and effective in repeated sampling experiments on simulated data of"three new"enterprises.
作者
张维群
成鹏东
ZHANG Weiqun;CHENG Pengdong(School of Statistics,Xi'an University of Finance and Economics,Xi'an 710100,China)
出处
《统计与信息论坛》
CSSCI
北大核心
2024年第3期3-13,共11页
Journal of Statistics and Information
基金
国家社会科学基金重点项目“基本公共服务均等化测度理论、方法与应用研究”(20ATJ002)
国家社会科学基金重大项目“国家治理能力现代化的测度理论、方法与进展评价”(21&ZD147)。
关键词
“三新”企业
抽样单元
样本分布结构
复合估计
自我加权设计
"three new"enterprises
sampling unit
sample distribution structure
compound estimation
self-weighted design
