人口普查多报及重报估计

Estimation for Erroneous Census Enumerations and Duplicates

人口普查中不可避免发生多报,包括重报和其他普查多报。针对当今一些国家的政府统计部门在普查多报估计中只估计普查多报或重报的问题,提出构建普查多报估计量及重报估计量的研究目标。为实现目标,采用文献解读、现场调查和抽样估计相结合的方法研究普查多报人口及重报人口的认定标准、重报人口的统计口径、多报人口的估计方法与样本多报人口数目的关系、多报及重报估计量及其抽样方差估计量的建立。使用美国明尼苏达大学人口中心建立的人口普查微观数据库进行实证分析。研究表明,估计的总体普查多报率和重报率在适当范围内,普查登记质量较高;泰勒扩张法适合于普查多报率及重报率的抽样方差计算。通过研究得出如下结论:为全面了解普查登记中的多报问题,除单独估计重报外,还要估计总的普查多报;认为普查多报估计不重要的观点,不仅与现实情况不吻合,还会干扰政府统计部门的普查多报估计工作。

The census inevitably produces erroneous census enumerations,including duplicates and other erroneous enumerations.The objective for constructing erroneous enumeration estimators and duplicate estimators are proposed,for the problem of some governmental statistical departments only estimate total erroneous enumerations or duplicates in erroneous enumeration estimations.In order to achieve the goal,literature interpretation,on-site investigation and sampling estimation are combined to study the identification standards for erroneous enumeration populations and duplicate populations,the statistical caliber of duplicate populations,relationships between the estimation methods and the amount of erroneous enumerations in sample,the construction of erroneous enumeration estimator and its sampling variance estimator based on the theory of finite population probability sampling and erroneous census enumeration estimation.An empirical analysis is conducted by census micro-database established by Population Center of Minnesota University.The results show that,the estimators of erroneous census enumeration rate and duplicate rate are proper with the quality of census registration.Taylor expansion method is suitable for calculating sampling variances to fully understand erroneous enumerations and duplicates enumerations in the census.It is also necessary to estimate the total erroneous enumerations besides the duplicate estimations,in order to gain a comprehensive understanding in census.The view of duplicate estimations in census doesn't matter is not only inconsistent with reality,but also interferes with the census works.