Measuring change in discrete latent variables is not the most common approach to the analysis of longitudinal data. However, there are recent examples of model developments that showcase why these types of models may be of interest. An early example is the latent transition model, which examines latent class membership over time. Another example is the Saltus model, an approach to capture discontinuous development that can be either applied to cross-sectional or, with appropriate extensions, to longitudinal data. Recent examples of more general approaches concerning multiple discrete latent variables can be found in recent literature on diagnostic classification models.
The presentation will compare and contrast discrete and continuous latent variable models for longitudinal approaches, discuss where predictions and model fit may differ, and will provide examples and illustrate fields of application.

Monan Professor in Education, Lynch School of Education, Boston College

Executive Director, TIMISS & PIRLS international Study Center, Boston College