Sample Size and Model Complexity: Impact and Implications for Variance Estimates in Two-Level Models

• In Event: Multilevel Modeling Methods and Applications II

Abstract

Much is still to be learned regarding small samples and hierarchical linear models (HLMs). Thus, in an effort to continue to expand our understanding of two-level HLMs under various sample size and other model conditions, Monte Carlo methods were used to examine convergence rates, non-positive G matrices, parameter point and interval estimates, RMSE, and both Type I error control and statistical power of tests associated with the variance parameters from two-level linear models. Outcomes were analyzed as a function of: level-1 sample size, level-2 sample size, intercept variance, slope variance, collinearity, and model complexity.

Authors

• Bethany A. Bell, University of South Carolina

• Jason Schoeneberger, Charlotte-Mecklenburg Schools

• Grant B. Morgan, Baylor University

• Min Zhu, University of South Carolina - Columbia

• Jeffrey D. Kromrey, University of South Florida

• John M. Ferron, University of South Florida