Comparison Between the Use of Composite Score and Latent Variable in Linear Growth Curve Modeling

• In Event: AERA Virtual Poster Session 2
  In Poster Session: Structural Equation Modeling

Abstract

A first-order linear growth curve modeling (1-LGCM) analyzes growth trajectories of a composite score, while second-order linear growth curve modeling (2-LGCM) examines those of a latent variable. The current study aims to provide an understanding of using composite score in LGCM, compared to the second-order latent variable. To achieve the goal, a simulation study will be conducted to examine statistical power and bias of growth parameter estimates from both 1-LGCM and 2-LGCM, conditional to the varying levels of measurement invariance, sample sizes, etc. Tentatively, as far as strict invariance was held, there was no difference in estimation between composite score and latent variable. However, we expect the difference in power and bias will occur when longitudinal invariance assumption was violated.

Authors

• Hahyeong Kim, University of Illinois at Urbana-Champaign

• Yan Xia, University of Illinois at Urbana-Champaign

• Xinchang Zhou, University of Illinois at Urbana-Champaign