Research On The Application Of Dual System Estimator In The Net Error Estimation Of The Census

After the census is completed,the quality of its enrolment is assessed.The usual method is to take a sample,estimate the actual number,and then subtract the census enrollment from the estimated value to obtain the net census error.Currently,dual-system estimators are the dominant method for estimating net census error.However,when constructing dual-system estimators,government statistics departments in China and many other countries do not stratify the overall population with equal probability;they do not use reasonable methods to deal with mobile population;they do not adopt scientific and effective methods to calculate the sampling of dual-system estimators.variance,the interaction bias of the dual-system estimator is not estimated,and so on.In response to the above problems,the research goal of dual-system estimators that conform to the sampling theory and the census quality assessment theory is clearly put forward.In order to achieve the goal,using the analysis methods of literature interpretation,comparative research,achievement reference and transplantation,the following contents were comprehensively and systematically studied: designing a post-event stratification scheme suitable for China;Capture” model;log-linear model of contingency table;sampling design,adopting stratified double sampling to draw probability samples with census plots as sampling units;constructing dual-system estimators within the equal probability population System estimator and comparative analysis of three methods for dealing with mobile population;sampling variance estimation of dual-system estimators;census net error estimation and sampling variance estimation;interaction bias of dual-system estimators.The research finds that:it is necessary to construct a dual-system estimator in the equal probability population layer;the population census and post-event count survey should be kept independent;the construction of a dual-system estimator based on mobile population with non-moving population and inward-moving population is suitable for practical work.When the census and post census surveys are negatively correlated,they will overestimate the overall actual number,and when there is a positive correlation,they will underestimate the overall actual number.The estimator of the dual-system estimator constructed by using the post-event count survey samples is relatively complicated,and the calculation of the sampling variance estimated by the knife-cut method is also very complicated.Therefore,it is very necessary to carry out empirical analysis,and use the empirical data to calculate the calculation process again.Innovation: Design a post-event stratification scheme suitable for the characteristics of China’s population census;compare and analyze the three methods of moving population processing,and construct a dual-system estimator suitable for population movement in my country.Research value: Combining theory and practice,carry out detailed and comprehensive analysis and research on the entire process of the construction of the dual-system estimator and its sampling variance estimator,especially the interaction bias of the dual-system estimator,and improve the dual-system estimation The relevant theory of quantity;the knife-cutting method in the multiple complex sampling of the primary unit knife-cutting technology and the adjustment technology when the basic unit sampling probability is not equal also has general theoretical guiding significance for research in other related fields;There are several deficiencies in the research and application of estimators,and a solution is proposed to provide a reference for my country’s 2030 population census net error estimation plan,which will help my country to formulate a more scientific population census plan and improve the accuracy of the population census.