Linear quantile regression models for longitudinal experiments: an overview

We provide an overview of linear quantile regression models for continuous responses repeatedly measured over time. We distinguish between marginal approaches, that explicitly model the data association structure, and conditional approaches, that consider individual-specific parameters to describe dependence among data and overdispersion. General estimation schemes are discussed and available software options are listed. We also mention methods to deal with non-ignorable missing values, with spatially dependent observations and nonparametric and semiparametric models. The paper is concluded by an overview of open issues in longitudinal quantile regression.