Speaker: Dr. Hao Wu, Boston College
Department of Psychology

Title: Identifying models in invariance testing for ordered categorical data.

Abstract: In this presentation I will consider the identification conditions of confirmatory factor analysis (CFA) models for ordered categorical outcomes with invariance of different types of parameters across groups. The usual practice of invariance testing is to first identify a baseline model with only invariance of the factor loading pattern and then test additional invariance of parameters based on this identified baseline model. A problem of this approach is that different ways to identify the baseline model may lead to different conclusions. In this work we give identification conditions for models with different levels of invariance by analyzing the reparametrizations that leave the observed category probabilities unchanged,. Tests based on this approach do not depend on the specific identification condition chosen for the baseline model.