A new variance component estimation for achieving multiple goals simultaneously


Session:

Confidentiality and Related Topics in SAE

Author: Masayo Hirose (Institute of Statistical Mathematics)
Abstract:

For the last several decades, area level models have played a critical role in the theory and practice of small area estimation. The implementation of an area level model does not require confidential micro data.  Aggregate statistics are modeled and thus the chance of disclosing information about a given individual is minimal. Relatively easier accessibility of aggregate statistics has helped researchers evaluating statistical methods and theory developed for area level models. For an area level model, we propose random effects variance estimator that simultaneously (i) improves on the estimation of the related shrinkage factors, (ii) protects empirical best linear unbiased predictors (EBLUP) of the random effects from the common over-shrinkage problem, (iii) avoids complex bias correction in generating strictly positive second-order unbiased mean square error (MSE) estimator either by the Taylor series or single parametric bootstrap method. The idea of achieving multiple desirable properties in an EBLUP method through a suitably devised random effects variance estimator is the first of its kind and holds promise in providing good inferences for random effects under the EBLUP framework. The proposed methodology is also evaluated using a Monte Carlo simulation study and real data analysis.