Research On Census Omission Estimation

The unmatched omissions estimator recommended by the United Nations Statistics Division to countries is currently the mainstream method for estimating census omission by government statistical offices in many countries.The disadvantage of this method is that it does not include the omission population that is missed in both the census and the the post quality survey,underestimating the overall census omission.In order to solve this problem,under the condition of comprehensive and in-depth interpretation of relevant domestic and foreign literature,the proposed research goal is to use the census population list,the post quality survey population list,the administrative record population list and its matching population list to construct the census omission synthetic estimator replace the unmatched omission estimator and other census omission estimators.In order to achieve this goal,literature possession and reading,achievement reference and transplantation,and sampling survey were used to study the following contents: missing unit omission estimator and linear missing unit omission estimator under three population lists;the synthetic estimator of census omission under two population movement scenarios;the synthetic estimator of census omission and its sampling variance estimator of equal probability population layer;the synthetic estimator of census omission and the estimator of sampling variance of the population;unmatched omission estimator;Comparison of synthetic estimator of census missing and unmatched omission estimator.The results show that compared with other census omission estimators,the synthetic estimation of census omission covers the entire population,is not limited by the independence of three population lists,and uses two auxiliary information of census omission list and administrative record population list,so the estimation accuracy is high.The synthetic estimator of census omission is a complex estimator,and its sampling variance cannot be calculated accurately,so it can only be approximated by stratified jackknife method.In approximating the sampling variance of the synthetic estimator of census omission,the covariance between the equal probability population layers should be calculated as well as the sampling variance of the equal probability population layers.Although the linear census omission estimate does not need to be established in the equal probability population layer,it is constructed in the equal probability population layer together with the missing unit census omission estimate for the convenience of calculation.Research innovation comes in two ways.First,the paper proposes the synthetic estimator of census omission,which solves the problem of underestimating the total census omission by using unmatched missing estimate,and promotes or enrichis the theory of census omission estimate to some extent.Secondly,the detailed calculation process of the synthetic missing estimator and its sampling variance is comprehensively demonstrated by using the microscopic data of the sample census area,which is helpful for readers to understand the synthetic estimator of census omission more easily.The research value lies in that the census omission synthetic estimator has a broad application prospect because of its theoretical advance and practical operability,which is helpful to the scientific formulation of the census omission estimation scheme by the government statistical department.