Comparison of Model Fit Evaluation Approaches in Multilevel Structural Equation Modeling

• In Event: Innovative Methods in Structural Equation Modeling

Abstract

The objective of this study is to compare the performance of model fit approaches in the MSEM. To achieve this goal, Monte Carlo simulation studies were carried out manipulating several research conditions: number of groups, group size, ICCs, group balance, and misspecification types. The results showed that (a) the level-specific approaches (Segregating approach and Partially-Saturated approach) performed better in detecting misspecification in the between-group level than the Simultaneous approach, (b) the ICC and group size influenced performance of model evaluation approaches, and (c) the Partially-Saturated approach is recommended when sample size is small or ICC is low.

Authors

• Song Jung, Seoul Educational Research & Information Institute

• Sehee Hong, Korea University