Bayesian Quantile Growth Curve Models for Longitudinal Data

Xin Tong, Dandan Tang

DOI: 10.64028/zhbi835956

Keywords: Quantile analysis, longitudinal data, Bayesian estimation, asymmetric Laplace distribution, growth curve modeling

Abstract: Longitudinal studies follow subjects across time, showing how subjects change and which factors are associated with interindividual variations in change. Despite its popularity, longitudinal research often faces methodological challenges. In this study, we introduce a robust Bayesian approach using conditional quantiles to address the nonnormality of data and population heterogeneity challenges in longitudinal studies. By converting the problem of estimating a quantile longitudinal model into a problem of obtaining the maximum likelihood estimator for a modified model with the assistance of the asymmetric Laplace distribution, Bayesian estimation methods can be conveniently used. Simulation studies have been conducted to evaluate the numerical performance of the quantile approach.

Submission Number: 26