Multilevel latent variable modeling : An application in educational testing
Jeroen Kornelis Vermunt · Data Archiving and Networked Services (DANS) · 2008
A framework for multilevel latent variable modeling is presented that includes many existing models as special cases. It is shown that parameters can be estimated by maximum likelihood using a special variant of the EM algorithm. An application is presented from the field of school effectiveness research. This application uses a novel multilevel mixture item response model which clusters schools based on the students’ latent abilities and the item difficulties.