Missing Data in Multilevel Structural Equation Models: A Multilevel Structural-After-Measurement Estimation Approach

• In Event: AERA Roundtable Session Sunday 8:00 am Four Seasons Ballroom 2-3
  In Roundtable Session: Applications and Advances in Structural Equation Modeling (Table 6)

Abstract

We develop a multilevel structural after measurement (MLSAM) estimator for structural equation models (SEMs) that accommodates missing data and probe the degree to which the method can retain key advantages that have been previously established. The results show that the proposed MLSAM missing data estimator emerges as a useful estimator when data are missing in small to moderate multilevel samples or in large multilevel samples with significant missing data. The proposed estimator outperforms conventional full information estimators in terms of convergence, bias, error variance, and power. The proposed estimator is implemented in R and are illustrated through a simple mediation example.

Authors

• Benjamin Kelcey, University of Cincinnati

• Fangxing Bai, Montana State University

• Amota Ataneka, University of Cincinnati