SOC 386L • Statistical Methods for Longitudinal Data Analysis
4:00 PM-5:30 PM
The objective of this course is to review the nature and illustrate the applicability of techniques for the analysis of longitudinal data. This course covers multiple regression models for data collected on the same subjects over time, as well as methods for analyzing event histories. The first half of the course provides an introduction to multilevel models for change (i.e., growth curve models), which are appropriate for the analysis of change in a continuous dependent variable over time. We will also review latent linear growth curve models, which use a structural equation modeling (SEM) approach. Growth curve models for categorical outcomes, as well as nonlinear growth curve models, will also be discussed. The second half of this course deals with event history analysis (i.e., survival analysis), which is a technique for modeling the transition from one status (or state) to another. Examples include life course transitions like marriage, birth, divorce, entry and exit from the labor force, etc. We will focus on discrete time and continuous time models that make few assumptions regarding time dependence of the hazard rate. As such, we will focus largely on semiparametric methods, such as the piecewise constant exponential and the Cox proportional hazard model. We will cover single transition as well as multi-state and competing risks methods. We will also consider multilevel hazard models and multiprocess models. Although it is impossible to depart even minimally from the contribution of other disciplines, the emphasis will be mainly on the analysis of relevant sociological and demographic phenomena.
J. D. Singer, and J. B. Willett (2003) Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York: Oxford University Press.