Professor Laura Dumitresu

Department of Mathematics, Washington University

Title : Robust small area estimation under semi-parametric mixed models
Abstract: In a small area framework empirical best linear unbiased predictors are efficient under a correct model specification but they might be highly influenced by the presence of outliers in the data, or departures from the normal distribution assumption of the random effects. We consider a semi-parametric regression model and using a mixed model approximation we develop a robust empirical best linear unbiased predictor of the small area mean and propose a conditional bootstrap estimator of the mean squared error. Results from a simulation study are presented.