The Impact of Ignoring Cross-Loadings on MIMIC’s (Multiple-Indicator, Multiple-Cause) Sensitivity in Detecting Uniform Measurement Noninvariance

• In Event: Sunday Roundtable Session 11:25 am
  In Roundtable Session: Measurement Issues in Structural Equation Modeling (Table 13)

Abstract

Multiple-indicator multiple-cause (MIMIC) modeling has been used frequently to test group mean difference on the latent factor while assuming factor loadings and item intercepts measurement invariance. Also, MIMIC has been used to test measurement invariance along with multi-group confirmatory factor analysis (CFA). Previous research about the performance of MIMIC modeling typically used one-factor model. When there are multiple factors and some nonignorable cross-loadings in the population model, the impact of ignoring the cross-loadings on its performance in detecting uniform measurement noninvariance in examined in this study. Preliminary results showed that when the noninvariant items happened to have cross-loadings that were ignored, the sensitivity of MIMIC decreased substantially.

Author

• Chunhua Cao, University of Alabama