Learn How To
Use basic multilevel models. Use three-level and cross-classified models. Use generalized multilevel models for discrete dependent variables.Who Should Attend
Researchers in psychology, education, social science, medicine, and business, or others analyzing data with multilevel nesting structure
Prerequisites
Before attending this course, you should:;
Preferably, be familiar with the basic structure and concepts of SAS (for example, the DATA step and procedures). Be familiar with concepts of linear models such as regression and ANOVA and with generalized linear models such as logistic regression. Be familiar with linear mixed models to enhance understanding, although this is not necessary to benefit from the course.;It is recommended that you complete SAS Programming 1: Essentials and Statistics 2: ANOVA and Regression or have equivalent knowledge before taking this course.SAS Products Covered
SAS/STAT
Course Outline
Introduction to Multilevel Models
Nested data structures. Ignoring dependence. Methods for modeling dependent data structures. The random-effects ANOVA model.Basic Multilevel ModelsRandom-effects regression. Centering predictors in multilevel models. Model building. A comment on notation (self-study). Intercepts as outcomes.Slopes as Outcomes and Model EvaluationSlopes as outcomes. Model assumptions. Model assessment and diagnostics. Maximum likelihood estimation.The Analysis of Repeated MeasuresThe conceptualization of a growth curve. The multilevel growth model. Time-invariant predictors of growth (self-study). Multiple groups models.Three-Level and Cross-Classified ModelsThree-level models. Three-level models with random slopes. Cross-classified models.Multilevel Models for Discrete Dependent VariablesDiscrete dependent variables. Generalized linear models. Multilevel generalized linear models. Additional considerations.Generalized Multilevel Linear Models for Longitudinal Data (Self-Study)Complexities of longitudinal data structures. The unconditional growth model for discrete dependent variables. Conditional growth models for discrete dependent variables.