[Forum SIS] Seminario Dimitris Rizopoulos

[Silvia Cagnone]

*********************************************************
             Annuncio seminario

*********************************************************

Dipartimento di Scienze Statistiche
Università di Bologna
via delle Belle Arti 41, 40126 Bologna


CHOOSING OPTIMAL WEIGHTS WHEN FITTING LATENT VARIABLE MODELS WITH THE PAIRWISE LIKELIHOOD

DIMITRIS RIZOPOULOS
University of Rotterdam

10 maggio 2011 ore 16:00  Aula III.



Abstract

Latent variable models are routinely used in many educational and sociological applications. In many case studies of this type it is typically required to postulate latent variable models that incorporate longitudinal effects.

 However, fitting such models with high-dimensional latent structures turns out to be a rather computationally challenging task. To this end and in order to decrease the computational burden several approaches have been recently proposed
in the literature that are based on composite likelihood ideas.

In this talk we will also follow this paradigm and particularly focus on the pairwise likelihood fitting procedure proposed by Fieuws and Verbeke (2006, Biometrics). This approach is based on fitting separately bivariate latent variable models for all possible pairs of time points, and then averaging parameters estimates and standard errors across pairs.

In this work we investigating alternative weighted average schemes that can improve the efficiency of the combined estimates in some settings.








Silvia Cagnone
Dipartimento di Scienze Statistiche "Paolo Fortunati"
Universita' di Bologna
Via delle belle arti 41 - 40126 Bologna, ITALY
Tel: +39 051 2098250 Fax: +39 051 232153
E-mail: silvia.cagnone at unibo.it
http://www2.stat.unibo.it/cagnone/

LA RICERCA C’È E SI VEDE:
5 per mille all'Università di Bologna - C.F.: 80007010376
http://www.unibo.it/Vademecum5permille.htm

Questa informativa è inserita in automatico dal sistema al fine esclusivo della realizzazione dei fini istituzionali dell’ente.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.stat.unipg.it/pipermail/sis/attachments/20110502/3be777db/attachment.html>