Empirical likelihood inference on the size of a closed population by modeling latent and observed heterogeneity

Francesco Bartolucci, University of Perugia

Co-authors: Antonio Forcina, University of Perugia

Abstract: For a flexible family of latent class models for capture-recapture data, we show how to make empirical likelihood (EL) inference as an alternative to the more common methods based on conditional or marginal likelihood functions. We focus in particular on the procedure for constructing profile likelihood confidence intervals for the population size. The advantages of the proposed models and estimation method are shown by a simulation study and two applications.