Nested and Clustered Data Structures: A Conceptual Overview

Odis Johnson Jr.

This seminar provides a conceptual overview of how to handle data that come from multiple units of analysis. An overview of hierarchical linear modeling will be provided, including how it extends from ANOVA, 2 and 3 level designs, HGLM, meta-analytic nested designs, cross-classified random effects modeling and how these approaches differ from multilevel regression. This session will also review approaches in creating variables for higher units of analysis from individual level data. Examples will refer to neighborhood-based analyses, policing, and schools. Information within this session will follow from the Sampling and Power Analysis Seminar and have applicability to randomized clustered designs. No software is required for this seminar.