Title: Bootstrap-based inference in generalized linear mixed models
Authors: Daniel Antonio Flores Agreda - Universite de Geneve (Switzerland) [presenting]
Eva Cantoni - University of Geneva (Switzerland)
Abstract: The focus is on two topics related to the inference of random effects using bootstrap methods. On a first stage we address the problem of uncertainty estimation in prediction for random effects in mixed models as measured by the Mean Squared Error of the Empirical Predictor (MSEP). We propose a non-parametric algorithm for estimation of the MSEP based on the Generalized Bootstrap for Estimating Equations adapted for the Linear Mixed Models setting. We apply this procedure in the framework of Generalized Linear Mixed Models and the EBP as we illustrate the properties of our proposal with simulation studies. On a second stage, we discuss extensions addressing the problem of random effect selection. We discuss the implementation of a penalized version of the Generalized Bootstrap for Estimating Equations with a re-parametrization that allows the construction of bootstrap confidence intervals for fixed and random effect parameters.