Rotation to Perfect Congruence and the Cross Validation of Component Weights Across Populations

Area of Science:

• Psychometrics
• Multivariate Statistics
• Component Analysis

Background:

• Congruence studies evaluate the recoverability of components (factors) from one population to another using the same variables.
• Assessing component stability and replicability across different datasets is crucial in statistical analysis.

Purpose of the Study:

• To analyze five key decisions in congruence studies.
• To introduce and advocate for a new strategy: rotation to perfect congruence.
• To compare the application of this strategy to variable-component correlations versus weights.

Main Methods:

• Detailed analysis of five decisions within congruence studies.
• Utilizing independent component analysis for confirmatory evidence.
• Employing oblique rotation to achieve perfect congruence.
• Applying the perfect congruence strategy to both variable-component correlations and component weights.

Main Results:

• Perfect congruence can always be attained through oblique rotation.
• Rotation to perfect congruence is proposed as a new strategy prioritizing explained variance.
• Applying the strategy to weights is preferred over variable-component correlations.
• Weight rotation serves as a cross-validation method, similar to the multiple group method.

Conclusions:

• The perfect congruence strategy, particularly when applied to weights, offers a valuable approach for assessing component recoverability.
• This method provides a practical and effective cross-validation technique.
• Results suggest superior practical performance when rotating weights to perfect congruence compared to rotating correlations.