A Two-Step Estimation Approach of Latent Class-Item Response Theory Models in the Presence of DIF and Multidimensionality
Estimation of Latent class item response theory (LC-IRT) models is typically performed using full-information maximum likelihood, which may be computationally demanding, susceptible to convergence problems, and subject to interpretational confounding. We propose a two-step estimation approach for unidimensional and mul...