|
Abstract:
|
Hierarchical data are common in many fields, from pharmaceuticals to agriculture to sociology. As you collect more and more data, information is likely to be observed on nested units at multiple levels, calling for the multilevel modeling approach. This tutorial will show you how to construct a multilevel model to account for variability at each level through both explanatory and random variables, in a way that shows the close relationship between multilevel models and mixed models. Then you will learn how to use the generalized linear mixed model procedure GLIMMIX in SAS/STAT® to estimate multilevel models for both continuous and discrete responses. You will see examples that illustrate the flexibility that PROC GLIMMIX offers for modeling within-cluster correlation, for disentangling multilevel explanatory variables, and for handling level-specific inferences. You will also learn about enhanced weighting options for PROC GLIMMIX that handle weights at different levels. Finally, you will see how to apply these features to analyzing complex survey data collected by multistage sampling with unequal sampling probabilities.
|
