An Empirical Comparison Of Two Minimum Residual Factor Extraction Methods.

Area of Science:

• Psychometrics
• Statistical analysis

Background:

• Factor analysis commonly uses diagonal communality estimates, which can be unknown.
• Comrey and Harman & Jones proposed methods using only off-diagonal elements to bypass this estimation.

Purpose of the Study:

• To empirically compare the Comrey and Harman & Jones factor analysis methods.
• To investigate the performance and outcomes of these alternative factor analysis techniques.

Main Methods:

• Factor analysis of a correlation matrix using only off-diagonal elements.
• Empirical comparison of the Comrey method and the Harman & Jones method.
• Analysis of derived communalities and rotated solutions.

Main Results:

• The Comrey method demonstrated significantly faster computation times.
• The Harman & Jones method resulted in higher derived communalities.
• Both methods yielded highly similar empirical rotated solutions.

Conclusions:

• Both factor analysis methods are viable alternatives when communality estimation is problematic.
• The choice between methods may depend on computational efficiency versus communality estimates.
• Further implications for factor analysis practices are discussed.