Clustered Data Mean You Need Multilevel Models, Right?

• In Event: 52.070 - Multilevel Modeling: Estimation, Model Selection, and Sample Size

Abstract

Multilevel models (MLMs) remain popular for modeling clustered data; however, other method exist that can model clustered data. Yet, a recent informal literature review found that MLMs are employed in published psychology studies 15 times more often than other clustered data methods. We explore a possible reason for MLMs being so dominant by comparing graduate school course syllabi from a variety of disciplines and then compare MLMs to generalized estimating equations, a straightforward method for modeling clustered data that may be a simpler alternative that more appropriately addresses researchers’ questions . A non-technical overview and applied examples are provided as is discussion of the merits and shortcomings of alternative methods for modeling clustered data.

Authors

• Daniel McNeish, University of Maryland

• Laura M. Stapleton, University of Maryland