Data structures are often hierarchical in that cases are clustered into groups (e.g., students in classrooms or repeated measures of individuals). Multilevel models are designed to model such nested data structures, where the usual assumption of independence of observations in classical techniques is violated, and within - and between - group effects may be separately estimated.
This short course provides an introduction to the basic concepts of linear multilevel models. Topics include approaches to analyzing nested data structures, computing and interpreting the intra-class correlation, level 1 predictors, partitioning variance into within- and between-group components, level 2 predictors, cross-level interactions, ML and REML estimation, model assumptions, model diagnostics and modeling longitudinal data structures.
The short course assumes familiarity with multiple linear regression, and will involve both lectures and data examples using SAS. Familiarity with the SAS environment is recommended.