[ABE-L] Seminário Conjunto UFSCar/ICMC – 24/04/2015 - 14h00 - no ICMC-USP

[Jorge Luis Bazan]

Esta mensagem épara divulgar a palestra desta semana do Seminário Conjunto doProgramaInterinstitucional de Pós-graduação em Estatística (PIPGEsICMC/USP e UFSCar), São Carlos.  Segue a informação.
LOCAL:AuditórioLuiz Antonio Fávaro(sala 4-111)– ICMC-USP,São Carlos, SP.
 TITULO:Bayesian longitudinal item responsemodeling with restricted covariance pattern structures
Palestrante:Caio Lucidius Naberezny Azevedo,IMECC-Unicamp (joint work with Dalton F. de Andrade, INE-UFSC andJean-Paul Fox, OMD-Univeristy of Twente, Holanda)
Resumo:Educationalstudies are often focused on growth in student performance andbackground variables that can explain developmental differencesacross examinees. To study educational progress, a flexible latentvariable model is required to model individual differences in growthgiven longitudinal item response data, while accounting fortime-heterogenous dependencies between measurements of studentperformance. Therefore, an item response theory model, to measuretime-specific latent traits, is extended to model growth using thelatent variable technology. The mean structure of the modelrepresents developmental change in student achievement. Restrictedcovariance pattern models are proposed to model thevariance-covariance structure of the student achievements. The mainadvantage of the extension is its ability to describe and explain thetype of time-heterogenous dependency between student achievements. Anefficient MCMC algorithm is given that can handle identificationrules and restricted parametric covariance structures. Areparameterization technique is used, where unrestricted modelparameters are sampled and transformed to obtain MCMC samples underthe implied restrictions.The study is motivated by a large-scalelongitudinal research program of the Brazilian Federal government toimprove the teaching quality and general structure of schools forprimary education. It is shown that the growth in math achievementcan be accurately measured when accounting for complex dependenciesover grades using time-heterogenous covariances structures.

Favordivulgar a possíveis interessados.

Saudações,Jorge


JorgeLuis BazánProfessorDoutor (MS-3)
Department of Applied Mathematics and Statistics,ICMC-USP
C.P. 668 - São Carlos, SP, Brazil - CEP 13560-970http://www.icmc.usp.br/pessoas/jlbazan/
-------------- Próxima Parte ----------
Um anexo em HTML foi limpo...
URL: <https://lists.ime.usp.br/archives/abe/attachments/20150420/f022f448/attachment.html>