Research On Registration Quality Evaluation For Population Census

The Population Census is carried out regularly in most countries and regions in the world.It can obtain demographic data to analyze the population situation and the change rule.The quality of population census is significative for population census.Population census evaluation has become an essential important part of the census in every country.The organization and implementation of the population census contain multiple stages,the error can be generated inevitably in each stage.Among that,it is the most difficult to control for the error in census registration stage,and had influence on the population census quality.Census registration quality is regarded as the primary target of population census evaluation in every country all over the world.Census omissions,census overcoverage and content error are the influence factor of the registration quality of population census.To this end,registration quality evaluation of population census is divided into two parts,coverage error evaluation and content error evaluation.Coverage error evaluation contains the population census net error,census overcoverage error and census omission error.And content error evaluation focuses on testing the consistency and difference of the demographic characteristics in the registration table.Coverage error is the emphasis of the census registration quality,because it concerns the estimation of the true population total.Post enumeration surveys,demographic analysis model,reverse record check,census overcoverage study and administration records are the commonly used methods for the registration quality evaluation,according to the practice experience of census registration quality evaluation in each country in the world.Among that,post enumeration surveys can simultaneously be applied to estimate census net error,census overcoverage and census omissions,and be widely used in evaluating the registration quality in many countries,such as United States,Switzerland,Australia,New Zealand,Uganda.National bureau of statistics of China uses post enumeration surveys to evaluate the census registration quality after population census,and constructs the dual system estimator to estimate the national total population based on capture-recapture model,and calculate the census net error.However,There are three drawbacks in constructing the dual system estimator.Firstly,post-stratification method is not used to meet the homogeneity assumption of the model.Secondly,population moving between the census and post enumeration is not be considered to meet the closed assumption.Thirdly,jackknife method is not used to calculate the variance of dual system estimator.Overall,the studies and work of census registration quality evaluation of China are inexperienced,and lack of solid basic theory research,have not form a scientific and reliable evaluation system of census registration quality,compared with some representative countries,such as United States.It is also need to be mention that,a lot of Chinese demographers make a deep research in the research on registration quality evaluation of population census.According to existing research results,however,there is still many problems to resolve.In the first place,these studies are more focus on adjusted register data for census,not the original register data.In the second place,the methods used in these studies are deviated from the mainstream methods of census registration quality evaluation used by national bureau of population census.In the thirdly place,these studies confuse some basic concept of census net error,census overcoverage and census omissions,caused the bewilderment of the theory research and index abuse of the practice.To these problems,the research object of this paper is interpreting the basic theory of population census registration quality evaluation,constructing the evaluation method system of Chinese population census registration quality based on the dual system estimator,and providing theoretical foundation for improving project design of Chinese population census in 2020 by studying and refer research achievement of census registration quality evaluation on the frontier of international,such as the guidelines and working papers of post enumeration surveys program of United States in 2000 and 2010 and Post Enumeration Surveys Operational Guidelines in United Nations.The main content of this article includes:(?)Researching and demonstrating the statistical nature of census net error.Net error is a statistical characteristic that describes the random distribution of census data.It respond the precision of census rather than the accuracy of census.(?)Summarizing a variety of indexes that represent different census registration status,it includes the concept of these indexes,constructing index system and the balance relationship of indexes.(?)Collecting and processing sample data,it includes design the address catalogue of housing units,matching the address catalogue between post enumeration surveys and population census,personal matching that ensured to be the matched housing units.(?)The statistical principle of capture-recapture,it includes background of experiment,estimator of capture-recapture model,unbiasedness of estimator,and application conditions of capture-recapture model.(V)Dual system estimator of registration quality evaluation of population census,it includes the respond relationship between dual system estimator and capture-recapture model,the unbiasedness conditions of dual system estimator,and the measurement of satisfied the unbiased conditions.(?)Applying post-stratification method for the population according to the homogeneity of registration probability,it includes the selection of stratification variable,dual system estimator under the condition of post-stratification.(?)Dual system estimator that reflecting the movement between census standard day and post enumeration surveys standard day.(?)Sampling of probability sample of finite population for the registration quality evaluation of population census,and how to use sample to constructing dual system estimator.(?)Estimation of census overcoverage,census omissions and content error.(?)Data simulation.It includes calculation of sampling weighting,dual system estimator of total population,jackknife variance estimator,and so on.