Comparison of Standardized Mean Difference Effect Size Estimates for Partially Nested Data

• In Event: Division D Graduate Student in Progress Research Gala

Abstract

The partially nested design is a special type of multilevel data where the treatment group entails clustering of individuals. A few approaches have been proposed to calculate the effect size for partially nested design data, including: an effect size based on use of summary statistics and another using parameter estimates from a multilevel model estimated using maximum likelihood estimation (MLE). In this study, we also introduce an effect size calculated using parameter estimates from a fixed-effects model. This simulation study compares these three effect size estimates in scenarios with a small number of clusters and a low intraclass correlation.

Author

• Young Ri Lee, The University of Texas at Austin