Testing for measurement invariance and latent mean differences across methods: interesting incremental information...

Area of Science:

• Psychological Measurement
• Quantitative Psychology
• Statistical Modeling

Background:

• Confirmatory factor analysis (CFA) is commonly used in multitrait-multimethod (MTMM) studies to assess convergent validity.
• Method effects can provide incremental information when mean structures and measurement invariance (MI) are incorporated into MTMM models.

Purpose of the Study:

• To present a modeling framework for testing MI in CFA-MTMM analyses.
• To discuss the relevance of MI in complex CFA-MTMM models with method factors.
• To examine whether MI is required for proper interpretation of trait or method factors across different models.

Main Methods:

• Focus on multiple-indicator CFA-MTMM models for structurally different methods (correlated traits-correlated (methods - 1), latent difference, latent means models) and interchangeable methods.
• Develop and present a modeling framework for testing MI as a preliminary step in CFA-MTMM analysis.
• Illustrate theoretical issues with an empirical application to an MTMM study of attention deficit and hyperactivity disorder (ADHD) using parent and teacher ratings.

Main Results:

• Some CFA-MTMM models inherently require or imply MI for valid interpretation of factors, while others do not.
• Testing for MI is critical in models with interchangeable methods to confirm their interchangeability.
• The inclusion of mean structures and MI tests enhances the understanding of method effects in MTMM investigations.

Conclusions:

• Incorporating mean structures and measurement invariance testing into CFA-MTMM models offers valuable insights beyond traditional analyses.
• The necessity of MI testing varies depending on the specific CFA-MTMM model employed.
• Proper interpretation of trait and method factors in MTMM studies is significantly influenced by the consideration of measurement invariance.