Journal of Official Statistics, Vol.7, No.1, 1991. pp. 57–67
Empirical Bayes Estimation of U.S. Undercount Based on Artificial Populations
Noel Cressie and Aref Dajani
Abstract:Estimators of undercount are difficult to assess and compare because true population counts are not available. Isaki et al. (1988) made the comparison by constructing an artificial population where “true” population counts were known. We show that the synthetic estimator they used is a special case of an empirical Bayes estimator of undercount, derived from a compound-distribution model for the undercount mechanism. The validity of this model, for the artificial population, can then be examined.
Keywords:Census counts; compound-distribution model; dual-system estimation; mean squared error; measures of improvement; post-enumeration survey; synthetic estimator; empirical Bayes estimator.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group