Longitudinal Data Analysis

3 Credits · Level: Graduate

This course intends to cover the major parametric and nonparametric models, estimation methods and inferences for the analysis of longitudinal or clustered data, i.e., repeated measurements data. The main topics include most of the well-known parametric, semiparametric and nonparametric regression models and their corresponding estimation and inference procedures. The regression models include the parametric marginal models, the linear and generalized linear mixed-effects models, the partially linear semiparametric models, and the structured nonparametric regression models. The estimation and inferences include the likelihood-based procedures, the kernel and basis approximation based nonparametric smoothing methods, the resampling subject bootstrap, and the asymptotically approximated inferences for longitudinal data. The practical aspects of the course will focus on the different longitudinal/clustered data structures, model construction and interpretations, computationally feasible estimation and inference methods, and applications to real longitudinal studies. Theoretical justifications of the estimation and inference methods will be outlined to show the unique techniques of asymptotic developments for repeated measurements data. But the detailed theoretical derivations will be assigned as reading materials. Students are expected to analyze some real datasets from biomedical studies using different models, conduct sensitivity studies and interpret the results. The main objective of the course is for students to build a solid background of the regression methods for longitudinal/clustered data and be able to apply these tools in practice. The R packages for longitudinal data and smoothing methods will be used throughout the course.