ABSTRACT
Erroneous enumerations in the census include multiple enumerations and other errors. The linear estimator for estimating these errors currently used in several countries leads to underestimation when the sample used for the estimation comprises few of these errors. The “ratio estimator” of the total number of erroneous enumerations overcomes this difficulty. This is the one used by China for the 2020 census. Empirical analysis shows that the ratio estimator provides a smaller sampling error than the linear estimator.
Acknowledgments
The authors thank two anonymous reviewers for their useful comments.