人口普查净覆盖误差估计

ESTIMATION FOR NET CENSUS COVERAGE ERROR

针对我国和其它许多国家的政府统计部门在应用双系统估计量时,忽视对总体人口等概率分层、未考虑人口移动,以及未采取科学有效的方法计算双系统估计量抽样方差的问题,提出建立基于捕获-再捕获模型的双系统估计量及分层刀切抽样方差估计量的研究目标。为实现目标,采取现场调查、抽样估计及文献解读相结合的方法研究全面登记和抽样登记的双系统估计量及其方差估计。研究发现,双系统估计量须在等概率人口层建立及使用,否则产生异质性偏差,低估总体实际人口数及净覆盖误差。研究创新在于,在对美国人口移动处理方法深入研究的基础上,构造出适合于在我国应用的人口移动的、抽样登记的双系统估计量。研究价值在于,所建立的双系统估计量有望应用于我国2030年人口普查净覆盖误差估计。

In view of the problems that the government statistical departments of our country and many other countries ignore the equal probability stratification of the overall population,do not consider population movement,and fail to adopt scientific and effective methods to calculate the sampling variance of dual-system estimator,when applying dual-system estimator,the research objective on constructing the dual-system estimator and the stratified Jack-Knife sampling variance estimator based on the capture-recapture model is proposed.In order to achieve the goal,a combination of field survey,sampling estimation and literature interpretation is adopted to study the dual-system estimator with complete enumerations and sampling enumerations and its variance estimation.The study finds that the dual-system estimator must be established and used in the equal-probability population stratum,otherwise a heterogeneous bias will occur,which will underestimate the overall actual population and net coverage error.The research innovation lies in the establishment of a dual-system estimator with population movement and sample enumerations,which is suitable for the application in our country based on the in-depth study of the US population movement processing methods.The value of the research is that the established dual-system estimator is expected to be applied to the net coverage error estimation of the 2030 census in our country.