Small area estimates for the fraction of the unemployed

Abstract

One of the main research trends in contemporary survey sampling and the need to improve the accuracy of the Lithuanian Labor force survey estimates in small geographic areas have stimulated this study. The aim of the paper is to compare area level models and estimation methods for the fraction of the unemployed using simulation based on the Lithuanian Labor Force Survey data. The Fay–Herriot area level model, estimated by empirical best linear unbiased prediction, and the unmatched logit-normal-normal and binomial- logit-normal models, estimated using hierarchical Bayes analysis, are applied. Bayesian imputation is used for areas without sample data. We suggest the composition of some model elements.