Multilevel Modeling: Introduction and Recent Advances for Behavioral Research

Abstract

Data in behavioral research usually follow a clustered structure, such as students nested in schools, participants nested in intervention groups, siblings within families, and, in longitudinal studies, repeated measures nested within persons. In this presentation, I will introduce the use of multilevel modeling (MLM) to obtain correct inferences with clustered data, and discuss new research questions that MLM can answer, such as group-specific (or person-specific) coefficients and the decomposition of individual-level and contextual effects. Practical recommendations will be provided on when MLM should be used. In addition, I will discuss some recent advances in MLM for behavioral research, including models for data with complex multilevel structures, effect size estimation, and multilevel bootstrapping.

Yuan Bo 袁博

Associate Professor of Psychology (Social Psychology)

My research examines the nature and dynamics of social norms, namely how norms may emerge and become stable, why norms may suddenly change, how is it possible that inefficient or unpopular norms survive, and what motivates people to obey norms. I combines laboratory and simulation experiments to test theoretical predictions and build empirically-grounded models of social norms and their dynamics.