Contenido del seminario
Programa de Doctorado en: Psicología Clínica y de la Salud
Línea/Tema de investigación: Modelos y Aplicaciones en Estadística y Psicometría
In IRT and CDM, we need to distinguish between structural (e.g., item parameters) and incidental (e.g., person-specific parameters). Previously, joint maximum likelihood estimation has been used to estimate the structural and incidental parameters simultaneously. Theoretically, this is not advisable. Neyman & Scott (1948) have shown that in the presence of incidental parameters, the maximum likelihood estimates of structural parameters may not be consistent. As a solution, instead of maximizing the likelihood function, which contains both structural and incidental parameters, we can work with the marginalized likelihood function, where the incidental parameters are integrated out. Alternatively, to bypass the calculation of (some of the) derivatives, we can use the EM Algorithm, which augments the observed data with latent data to simplify calculations, and make the subsequent maximization easier. Estimation of the 1PL and 2PL model parameters will be discussed in this session.
About the speaker:
Jimmy de la Torre is a Professor in the Faculty of Education at The University of Hong Kong. He is also currently a Chair Professor at the National Taichung University of Education in Taiwan, and an Honorary Professor at the Universidad Autonoma de Madrid in Spain. His primary research interests are in the field of psychological and educational testing and measurement, and the use of diagnostic assessment to support classroom teaching and learning. As one of the leading researchers in the field of cognitive diagnosis modeling, his work has covered both theoretical and implementation issues in this area. In 2009, he was named by the White House as one of the recipients of the Presidential Early Career Awards for Scientists and Engineers. He also received the Jason Millman Promising Measurement Scholar Award in 2009 from the National Council on Measurement in Education. He is the editor-in-chief of the Journal of Educational Measurement, an associate editor of Applied Psychological Measurement, and a member of the Psychometric Society Board of Trustees.