Linear Confirmatory Factor Models to Evaluate Multitrait-Multimethod Matrices: The Effects of Number of Indicators...

Area of Science:

• Psychometrics
• Quantitative Psychology
• Statistical Modeling

Background:

• Multitrait-multimethod (MTMM) data analysis is crucial for understanding construct validity.
• Confirmatory factor analysis (CFA) offers models to analyze MTMM data, including correlated traits-correlated methods (CTCM) and correlated traits-correlated uniqueness (CTCU) models.
• Understanding the performance of these models under different conditions is essential for accurate data interpretation.

Purpose of the Study:

• To evaluate the performance of the CTCM and CTCU models in confirmatory factor analysis of MTMM data.
• To assess model bias and accuracy of estimates through Monte Carlo simulations.
• To provide recommendations for choosing between CTCM and CTCU models based on data characteristics.

Main Methods:

• Two Monte Carlo simulation studies were conducted with 100 replications per condition.
• Study one examined MTMM designs with a single indicator per trait-method combination.
• Study two investigated MTMM designs with two or more indicators per trait-method combination.

Main Results:

• The CTCU model demonstrated superior performance and accuracy with single indicators per trait-method combination.
• The CTCM model showed better performance and accuracy when two or more indicators per trait-method combination were utilized.
• Model estimates were generally better when method factors were orthogonal, even for the CTCM model.

Conclusions:

• The CTCU model is recommended for MTMM analyses with a single indicator per trait-method combination.
• The CTCM model is recommended for MTMM analyses with multiple indicators per trait-method combination.
• Researchers should carefully consider the number of indicators per trait-method when selecting between CTCM and CTCU models for MTMM data analysis.