A0479
Title: Multilevel latent class analysis: Stepwise estimation
Authors: Zsuzsa Bakk - Leiden university (Netherlands) [presenting]
Abstract: A two-step estimator is proposed for multilevel latent class analysis with co-variates that separates the estimation of the measurement and structural model. Keeping the measurement model fixed in step two, when covariates are added to the model, it is possible to obtain an unbiased and efficient stepwise estimator. We investigate the bias and the efficiency of the proposed estimator via a simulation study. The results show that the proposed two-step estimator is less biased than the alternative three-step estimator and almost as efficient as the one-step estimator.
