Structural After Measurement Estimation of Three-Level Structural Equation Models

• In Event: Estimation Methods for Multilevel Models

Abstract

A major limitation of estimation in multilevel structural equation modeling (MLSEMing) is that they often fail to converge or incur significant bias when implemented in studies with a small to moderate sample in three-level structures (e.g., fewer than 100 schools with 4 classrooms and 20 students each). To address similar limitations in single- and two-level SEM, recent literature has developed a bias-corrected “structural after measurement” (SAM) estimator that converges more regularly than full information estimators and delivers consistent parameter estimates even with typical but small to moderate sample sizes. We derive extensions to this framework for three-level SEMs and probe the degree to which the estimator can retain these advantages with small to moderate three-level samples.

Authors

• Benjamin Kelcey, University of Cincinnati

• Kyle T. Cox, University of North Carolina - Charlotte

• Fangxing Bai, University of Cincinnati