Relatore Prof.ssa Silvia Cagnone - Dipartimento di Scienze Statistiche – Università di Bologna
Data 27 ottobre 2023 ore 12:00
Luogo Aula 202 Dipartimento di Economia
Abstract
Latent variable models are a powerful tool across various research domains, when the constructs of interest are not directly observable. However, estimating these models through likelihood-based methods can become challenging, especially when dealing with numerous continuous latent variables and random effects, as the integrals within the likelihood function lack analytical solutions. To address this issue, in the literature, several approaches have been proposed. Notably, the pairwise likelihood method and dimension-wise quadrature have emerged as effective solutions, yielding estimators with desirable properties.
In this study, we undertake a comparative analysis of the unweighted and weighted versions of the pairwise likelihood method against dimension-wise quadrature in the context of generalized linear latent variable models for multivariate panel data.
This is a joint work with Silvia Bianconcini, Department of Statistical Sciences, University of Bologna
Organizzazione Silvia Pandolfi (silvia.pandolfi@unipg.it)
Locandina (in formato pdf)