Small area estimation of general parameters under a non-informative sample design

Tuesday, April 21, 2020 - 12:30pm to 1:30pm
Event Type: 

Speaker: Yanghyeon Cho

Abstract: The main problems in small area estimation are how to predict the area quantities of interest and assess the prediction errors. In this talk, we will illustrate the empirical best prediction (EBP) method proposed by Molina and Rao (2010), which can be applied to predicting both linear and nonlinear parameters. For the prediction error problem, multiple estimators of mean squared error (MSE) will be introduced and compared under the non-informative sample design. Furthermore, we will show how the adjusted likelihood estimator can be used to solve the singular fits of model parameters estimation and how much the improvement was in the simulation study. Based on these studies, we will suggest our future work.