Using hierarchical linear modeling to analyze grouped data.

This article discusses how to use a random coefficient modeling technique known as hierarchical linear modeling to analyze data collected within groups. The article describes how to use this technique to examine group- and individual-level phenomena, including examination of how individual-level relationships vary as a function of group characteristics. A comparison of hierarchical linear modeling with more traditional, ordinary-least-squares techniques and a presentation of how to implement analyses to test specific hypotheses are included. Included are brief discussions of pertinent issues such as the impact of different centering options, the analysis of categorical variables, distinctions between random and fixed effects, and balanced and unbalanced designs. (PsycINFO Database Record (c) 2010 APA, all rights reserved)