Longitudinal data consists of repeated measurements of the same individuals or experimental units over time. Analysis of longitudinal data can be challenging because repeated observations from the same subject are often temporally autocorrelated, violating the independence assumption of standard regression models. To deal with this issue, regression models can be extended to include covariance structures that model the dependence among observations. This workshop will give a practical guide to incorporating covariance structures in regression models for longitudinal data, with worked examples in R (SAS and Stata code will also be provided). Topics covered include:
- Checking for temporal autocorrelation
- Covariance structures for equally spaced and unequally spaced time series
- Including covariance structures in linear models, mixed models, and GLMs
- Selecting the right covariance structure
This workshop assumes familiarity with linear mixed effects models.