Multilevel Models for Longitudinal Data Using SAS

Instructor(s):

• Lesa Hoffman, Psychology, University of Nebraska-Lincoln

Multilevel models are known by many synonyms (hierarchical linear models, general linear mixed models). The defining feature of these models is their capacity to provide quantification and prediction of random variance due to multiple sampling dimensions (across occasions, persons, or groups). Multilevel models offer many advantages for analyzing longitudinal data, such as flexibility strategies for modeling patterns of variance and covariance over time (alternative covariance structures or random effects), the possibility of examining time-invariant or time-varying predictor effects, and the use of all available complete observations. This workshop will serve as an applied introduction to multilevel models for longitudinal data and extensions thereof. The first day will be spent reviewing general linear models and then introducing the multilevel model, as well as the necessary data transformations it requires. The second day will be spent fitting unconditional growth models and on the rules of model comparisons. The third day will be spent examining time-invariant and time-varying predictors. The fourth day will be spent on two-level and three-level extensions of the multilevel model for clustered longitudinal data. The fifth day will be spent on multivariate outcomes and other special topics. The course will include hands-on practice using SAS PROC MIXED, but analysis scripts for SPSS MIXED and Mplus will also be provided.

Dates:  June 22-26 

Location:  University of North Carolina, Chapel Hill

Co-sponsor: The Odum Institute for Research in Social Science

Fee:  Member: $1800; Non-member: $4000

This course is limited to 20 participants.