A Multilevel Model Primer Using SAS PROC MIXED

Abstract

This paper provides an introduction to specifying multilevel models using PROC MIXED. After a brief introduction to the field of multilevel modeling, users are provided with concrete examples of how PROC MIXED can be used to estimate (a) two-level organizational models, (b) two-level growth models, and (c) three-level organizational models. Both random intercept and random intercept and slope models are illustrated. Examples are shown using different real world data sources, including the publically available Early Childhood Longitudinal Study–Kindergarten cohort data. For each example, different research questions are examined through both narrative explanations and examples of the PROC MIXED code and corresponding output.