Testing for Measurement Invariance With Ordinal Data: A Robust Weighted Least Squares Approach

• In Event: Poster Session 3
  In Poster Session: Division D Section 2 Quantitative Methods and Statistical Theory

Abstract

In the social and behavioral sciences, a robust weighted least squares approach has been widely used in the confirmatory factor analytic model when observed variables are ordinal. Testing for measurement invariance across group was carried out using multiple-group CFA models in the study. This study first validated the factor structure of the Peer Victimization Scale (PVS) by CFA and tested for measurement invariance on the PVS across gender. This procedure was illustrated on an empirical data from the Child Development Supplement (CDS) datasets with 440 elementary school students. Results showed (1) the single factor structure of the PVS was invariant across gender, and (2) no latent mean difference on peer victimization between boys and girls was detected.

Author

• Cheng-Hsien Li, Michigan State University