STATISTICAL AND PSYCHOMETRIC INFERENCE IN PRINCIPAL COMPONENTS ANALYSIS

Area of Science:

• Statistics
• Psychometrics
• Multivariate Analysis

Background:

• Principal Components Analysis (PCA) relies on statistical inference for correlation matrix estimation.
• Psychometric inference assesses internal consistency of components for retention decisions.

Purpose of the Study:

• To propose a modified statistical inference approach for PCA.
• To extend this approach for estimating population component variance, true score variance, and reliability.
• To enable statistically robust psychometric significance testing.

Main Methods:

• Modified statistical inference focusing on population component variance.
• Estimation of true score variance and component reliabilities.
• Application to psychometric significance testing in PCA.

Main Results:

• A unified framework for statistical and psychometric inference in PCA.
• Improved criteria for component retention based on reliability estimates.
• Statistical grounding for psychometric significance tests.

Conclusions:

• The proposed approach enhances the rigor of PCA by integrating statistical and psychometric inference.
• This method provides a more reliable basis for component selection and interpretation.
• It advances the statistical nature of psychometric significance testing.