On Small area estimation of poverty proportions under unit‐level logit mixed models

SEMINARIO DEL DOCTORADO DE ESTADÍSTICA, OPTIMIZACIÓN Y MATEMÁTICA APLICADA. CURSO 2015-2016

Título: On Small area estimation of poverty proportions under unit‐level logit mixed models

Ponentes: Prof. Dr. Tomas Hobza (Universidad Politécnica Checa en Praga, República Checa)

Fecha: martes 12 de julio de 2016 a las 12:00 horas

Lugar: Sala de Seminarios del Instituto Universitario de Investigación CIO, Edificio Torretamarit

Resumen:

This contribution deals with empirical best predictors (EBPs) and plug‐in estimators of small area proportions under a unit‐level logit mixed models. In the first part of the talk a model without time effects is presented. Behavior of the EBPs based on parameter estimates obtained by the method of simulated moments is studied by a Monte‐Carlo simulation experiment. Performance of an estimator of the mean squared error of EBPs based on approximation of the analytic form is compared with performance of a bootstrap estimator. In the second part of the talk we propose a more general temporal model which borrows strength from the past for better estimating the present. The maximum likelihood estimates of the model parameters are obtained by maximizing the Laplace approximation to the model log‐likelihood. The behavior of the introduced EBPs and plug‐in estimators is again studied by a simulation experiment. A procedure for bootstrap estimation of the mean squared errors is proposed. Finally, an application of the studied EBPs to data from the 2012 and 2013 Spanish living conditions surveys is presented for both models. The target is the estimation of proportions of people under the poverty line in counties of the region of Valencia.

Breve Bio:

Tomáš Hobza currently is Assistant Professor at the Czech Technical University in Prague, Faculty of Nuclear Sciences and Physical Engineering (CTU FNSPE), Department of Mathematics. Since 1999, he has taught courses on Combinatorics and Probability, Mathematical Statistics, Applied Statistical Methods, Information Theory and Generalized Linear Models. He has participated in more than nine research projects and has published more than 30 papers in the field of statistical methodology. His main actual research interest is on the field of Small Area Estimation, where he is developing statistical tools for estimating poverty and labour market indicators in territories where the survey sample sizes are not large enough for obtaining reliable direct estimates.