The Role of Sampling Ratio in Multilevel Structural Equation Models

• In Event: Methods for Multilevel, Categorical, and Multidimensional Data

Abstract

Multilevel structural equation (MSEM) models allow researchers to consider latent factor structures at multiple levels simultaneously by decomposing within- and between-group variation. The role of the sampling ratio in MSEM models remains an area understudied in the current literature. This paper explores the role of the sampling ratio in multilevel measurement models with a focus on bias and consistency in latent L2 construct variance estimates through a Monte Carlo simulation study. Results suggest the sampling ratio is not related to convergence rates and bias or precision in L2 construct estimates. Findings from this study may have important and relevant implications for both quantitative methodologists and applied researchers alike.

Authors

• Joseph M. Kush, James Madison University

• Timothy R. Konold, University of Virginia

• Catherine P. Bradshaw, University of Virginia