The Effects of Overextraction on Factor and Component Analysis.

Area of Science:

• Psychometrics
• Statistical Analysis

Background:

• Principal Component Analysis (PCA) and Maximum Likelihood Factor Analysis (MLFA) are common methods for data reduction.
• Overextraction of factors or components can occur, potentially affecting the validity of results.
• Understanding the impact of overextraction is crucial for accurate data interpretation.

Purpose of the Study:

• To examine the effects of overextracting factors and components within and between MLFA and PCA.
• To assess how varying levels of saturation, sample size, and variable-to-component ratios influence score reproduction under overextraction.

Main Methods:

• Computer-simulated datasets were generated with manipulated saturation, sample size, and variable-to-component ratios.
• Study 1: Correlated scores based on incorrect vs. correct patterns within each method (MLFA, PCA) after overextraction.
• Study 2: Correlated scores between MLFA and PCA after overextraction.

Main Results:

• Overextraction negatively affected scores, but strong component and factor patterns demonstrated robustness.
• Low item saturation and small sample sizes led to degraded score reproduction, especially when combined.
• Component and factor scores remained highly correlated even with maximal overextraction; dissimilarity was greatest under low saturation and sample size conditions.

Conclusions:

• Researchers should be cautious about overextracting factors/components in MLFA and PCA.
• Strong factor patterns offer some resilience to overextraction, but low saturation and sample size exacerbate score degradation.
• Guidelines and cautions are provided for interpreting results affected by overextraction in factor analysis and PCA.