Latent autoregressive models for the examination of self-rated health trends over time

What is it about?

A new formulation of the latent autoregressive model for longitudinal data is proposed by considering mixture models in order to account for binary, ordinal or continuous variables. The proposal is illustrated by considering the self-reported health status and its association with covariates like gender, race, education, and age.

Featured Image

Why is it important?

The proposed model is much more flexible compared to the other models in the literature for longitudinal data based on a continuous latent process. We perform maximum likelihood estimation of the model parameters. We disentangle three main groups of individuals by considering a sample from the Health and Retirement Study and we predict their individual effects over time. The applied example is important since self-rated health is predictor of morbidity, healthcare utilization, hospitalization and mortality.

Perspectives