What Is Exploratory Factor Analysis?
Exploratory Factor Analysis (EFA) identifies the underlying factor structure among a set of observed variables. It reduces many variables into fewer latent factors based on shared variance. EFA is the cornerstone of scale development and construct validity studies — used when the factor structure is not known in advance.
Prerequisites: KMO and Bartlett's Test
- KMO (Kaiser-Meyer-Olkin): Values ≥0.70 are acceptable, ≥0.80 are good, ≥0.90 are excellent. Values below 0.60 suggest factor analysis is inappropriate.
- Bartlett's Test of Sphericity: Must be significant (p<0.05), confirming that the correlation matrix is not an identity matrix.
Running EFA in SPSS
Navigate to Analyze → Dimension Reduction → Factor.
- Move all items to the Variables box.
- Extraction: Select Principal Axis Factoring; eigenvalue criterion ≥1.0.
- Rotation: Use Varimax (orthogonal) if factors are expected to be uncorrelated; use Promax (oblique) if correlated factors are expected.
- Options: Suppress coefficients below 0.30 or 0.40 to improve readability.
Interpreting the Output
Total Variance Explained: Cumulative variance explained by retained factors should ideally exceed 50%. Rotated Factor Matrix: Each item's loading on each factor. Items with loadings ≥0.40 on a factor belong to that factor.
Cross-Loading Problem
If an item loads ≥0.40 on two or more factors, it has a cross-loading problem. Such items are typically removed and the analysis is re-run. Continue until a clean, interpretable factor structure is achieved.
Naming Your Factors
Examine the items within each factor and assign a conceptual label that best represents what they share. This step is interpretive and requires domain expertise.
APA Reporting
Report: KMO value, Bartlett's test result, total variance explained, number of factors retained, and a table showing factor loadings for all items. Boss Statistics handles the entire EFA workflow for you.