A0672
Title: Capturing static and dynamic heterogeneity in a Hidden Markov model for binary data
Authors: Dalila Failli - University of Perugia (Italy) [presenting]
Maria Francesca Marino - University of Florence (Italy)
Francesca Martella - La Sapienza University of Rome (Italy)
Abstract: A model-based clustering approach for multivariate binary longitudinal data that accounts for both time-varying and time-constant sources of unobserved heterogeneity. Specifically, a latent Markov model (LMM) specification is considered to capture the dynamic nature of the data and enable dynamic clustering of units over time. Additionally, a multidimensional continuous latent trait is incorporated to capture residual time-constant heterogeneity among units within the same latent state at any given time point, in terms of their responses to the multivariate binary variables. For parameter estimation, the standard Baum-Welch algorithm is extended to accommodate the presence of the continuous latent trait. In this context, the solution of multidimensional integrals not available in closed form is required. A variational approximation is considered to overcome the issue. The performance of the proposed approach is evaluated in terms of parameter recovery and clustering accuracy. The ability of different model selection procedures in identifying the optimal number of latent states and latent trait dimensions is also evaluated.
