Small Area Estimation with Covariates Perturbed for Disclosure Limitation

We exploit the connections between measurement error and data perturbation for disclosure limitation in the context of small area estimation. Our starting point is an area level model in which some of the covariates (all continuous) are measured with error. Using a fully Bayesian approach, we extend such model including continuous and categorical auxiliary variables, both perturbed by disclosure limitation methods, with masking distributions fixed according to the assumed protection mechanism. In order to investigate the feasibility of the proposed method, we conduct an extensive simulation study exploring the effect of different protection scenarios on the small area mean predictions. We also perform a comparative analysis of the proposed estimator.