Some Mathematical Properties of the Matrix Decomposition Solution in Factor Analysis.

Area of Science:

• Statistics
• Psychometrics
• Data Analysis

Background:

• Factor analysis (FA) is a statistical method used to describe variability among observed, correlated variables in terms of a potentially lower number of unobserved variables called factors.
• Traditional FA models involve estimating parameters like factor loadings and unique variances.
• Matrix decomposition factor analysis (MDFA) presents a recent alternative framework.

Purpose of the Study:

• To fully investigate the properties of the Matrix Decomposition Factor Analysis (MDFA) procedure.
• To discover new properties and provide more explicit derivations of existing ones.
• To assess the uniqueness of results, factor covariances, and factor score indeterminacy within the MDFA framework.

Main Methods:

• The study treats all factor analysis model parameters as fixed unknown matrices, simplifying the model to a data matrix decomposition.
• MDFA parameters are determined by minimizing the discrepancy between the observed data and the decomposed model.
• The research explores mathematical properties related to uniqueness, covariances, and indeterminacy, illustrated with a real data example.

Main Results:

• Several new properties of MDFA are identified and explored.
• Existing properties are derived with greater explicitness, offering deeper insights.
• The study assesses the uniqueness of solutions, covariances among factors and residuals, and the indeterminacy of factor scores.

Conclusions:

• MDFA represents a distinct approach to factor analysis with unique mathematical properties.
• The findings contribute to a more comprehensive understanding of MDFA's theoretical underpinnings and practical implications.
• The study validates MDFA's utility through the analysis of a real-world dataset.