中文摘要：

本文针对作为当今人口普查质量评估领域主流方法的双系统估计量因受独立性不满足困扰而有偏估计总体人数的问题,明确提出了用三系统估计量替代双系统估计量的研究目标。三系统估计量指的是依据对同一时点上同一总体的人口普查、人口普查之后的质量评估调查和人口行政记录系统这三个不同途径获得的三份人口登记数据来估计总体人数所使用的估计量。为实现上述目标,采用实地调研、数理模型、抽样估计与实证分析相结合的研究方法,以及借鉴国际相关前沿研究成果的研究思路,研究了人口普查质量评估中三系统估计量的构建、三系统估计量应用操作等问题。通过理论与实证研究,得出如下结论：三系统估计量能够有效地摆脱双系统估计量的独立性约束从而解决估计量偏误的问题;将三系统估计量应用于人口普查质量评估,必须科学合理地解决三个人口登记系统定格在统一的标准时点、人口在三个登记系统等概率分层、用人口普查质量评估调查样本对三系统估计量再估计等实践方面的问题;基于Logistic回归模型的三系统估计量可以较好地实现人口等概率分层目标,但尚有较多的理论与实践问题有待研究解决,在这一问题上有着相当大的研究空间。建议我国在未来人口普查质量评估中应用三系统估计量。

英文摘要：

This paper clearly puts forward the research goal that triple system replaces dual system estimator aimed at the question about the population count is biased estimated by dual system estimator originated from independence failure, which is the mainstream method in the field of population quality assessment. Triple system estimator refers to the population count estimator that is constructed based on three population registration lists from the census, quality assessment survey after the census and administrative record system, which belongs to the same time and the same population. In order to achieve the above goal, in the paper, the research methods of the field investigation, mathematical model, sampling estimation and empirical analysis are adopted, and absorbing international advanced related research result idea. Based on this idea, this paper studies the construction of the triple system estimator in the nualitv evaluation of the census and the application and so on. Through the theoretical and empirical research, the following conclusions are drawn： triple system estimator can effectively get rid of the independent constraints of the dual system estimator to solve the problem of the biased estimator; in order to apply triple system estimator to the census quality assessment, it is necessary to scientifically and reasonably solve some questions that three population registration systems must be in a unified standard time ; each person has the same registration probability in any of three registration systems ; estimator of triple system estimator must be constructed with quality assessment survey ; although triple system estimator based on Logistic regression model can be better used to achieve the equal probability stratified target there are still many theoretical and practical problems to be solved, there is a considerable research space on this issue. It is suggested that the National Bureau of statistics of China should apply triple system estimator in the quality evaluation of future population ce