Standardized Mean Differences in Two-Level Partially Nested Models

• In Event: Multilevel Modeling: Nonnested Multilevel Data, Measurement Invariance, and Nonlinear Models

Abstract

The present paper discussed two methods to obtain standardized mean difference (D) effect size and the corresponding sampling variance for partially nested cluster randomized designs. The first method requires input of summary statistics such as observed means, variances, and intraclass correlation, and would be useful for meta-analysts and secondary data analysts. The second method takes estimated variance components as input and would be of interest for primary researchers. Real data are used to demonstrate the method. Furthermore, bias on D due to incorrect modelling of partially nested data is shown to increase with larger intraclass correlation and cluster size.

Authors

• Hok Chio Lai, Texas A&M University - College Station

• Oiman Kwok, Texas A&M University - College Station