Small Sample Bias Correction in Multilevel Meta-Analyses of Standardized Single-Case Experimental Data

• In Event: 25.026 - Challenges in Meta-Analysis

Abstract

Given the possibility of bias in effect size estimations of multilevel meta-analysis of standardized single-case data, several adjustments were evaluated in a simulation study. The results confirmed the existence of bias when estimating an overall effect based on standardized data, as well as bias of corresponding standard error estimates. The relative bias decreases as the number of measurement occasions increases. The mean squared error (MSE) is smaller as the number of measurement occasions and the number of studies increase. Hedges’ bias correction procedure performed better in a variety of conditions. As the number of measurement occasions and the number of studies increase, the relative bias of standard errors decreases significantly. In general, Hedges’ bias correction procedure led to better estimations.

Authors

• Laleh Jamshidi, KU Leuven

• John M. Ferron, University of South Florida

• Mariola Moeyaert, University at Albany

• Tasha Beretvas, The University of Texas at Austin

• Wim Van den Noortgate, KU Leuven