New Publication: Small Area Estimation Using Area-Level Linked Data

Read about Small Area Estimation Using Area-Level Linked Data published in Journal of Survey Statistics and Methodology by Enrico Fabrizi, Nicola Salvati and Ray Chambers.
Abstract:
Official statistics increasingly use linked survey and administrative data for small area estimation, but record linkage error can distort direct estimators in area-level Fay–Herriot models. We develop a likelihood-based framework for such models under an Exchangeable Linkage Error mechanism, using the Missing Information Principle to obtain linkage error-adjusted maximum likelihood estimators of regression coefficients and variance components, an empirical best predictor of true area means, and a mean squared error estimator. Simulation results show that ignoring linkage error leads to substantial bias and loss of efficiency, while the proposed methods restore unbiasedness and provide accurate uncertainty quantification. An application to estimation of average equivalized income for 113 labor market areas in Central Italy, based on European Union Statistics on Income and Living Conditions (EU-SILC) data linked to administrative registers, illustrates the practical benefits of the approach.


