Title
Longitudinal Mixed Membership Trajectory Models for Survey Data on Disability
Speaker
Daniel Manrique-Vallier
Abstract
I introduce a new family of mixed-membership models to analyze longitudinal data that assume the existence of a small number of “typical” or “extreme” classes of individuals and model their evolution over time. These models regard individuals as belonging to all of these classes in different degree, by considering them as convex weighted combinations of the extreme classes. This way, they describe distinct general tendencies (the extreme cases) while accounting for the individual variability. I propose a full Bayesian specification and estimation methods based on Markov Chain Monte Carlo sampling and apply the method to data from the National Long Term Care Survey (NLTCS), a longitudinal survey with six completed waves aimed to assess the state and characteristics of disability among U.S. citizens age 65 and above. I combine this mixed-membership latent trajectory approach with a survival model that incorporates information on deaths and propose a simple extension that enables us to answer questions about changes across generations.
