Day 34: Exploratory Factor Analysis (EFA) in SPSS – Identifying Underlying Constructs
Welcome to Day 34 of your 50-day SPSS learning journey! Today, we’ll explore Exploratory Factor Analysis (EFA), a technique used to identify latent constructs (hidden variables) that explain patterns in your data. EFA is commonly used in survey research, psychology, social sciences, and marketing to group related items into meaningful factors.
What is Exploratory Factor Analysis (EFA)?
Exploratory Factor Analysis (EFA) is a statistical method used to:
- Reduce many observed variables into a smaller number of underlying factors.
- Identify relationships among variables and detect latent constructs.
- Determine which variables group together, forming meaningful dimensions.
For example:
- In a customer satisfaction survey, EFA can group items into Service Quality, Product Value, and Customer Support.
- In psychological research, EFA can reveal factors like Extraversion, Conscientiousness, and Neuroticism from personality traits.
When to Use EFA?
Use Exploratory Factor Analysis when:
- You suspect underlying factors exist but don’t know their exact structure.
- You need to reduce the number of variables for further analysis.
- Your dataset contains correlated variables measuring related concepts.
Key Assumptions of EFA
- Adequate Sample Size: At least 5-10 cases per variable for reliable results.
- Multicollinearity: Variables should be correlated but not perfectly correlated.
- Kaiser-Meyer-Olkin (KMO) Test: Measures sampling adequacy (values above 0.6 are acceptable).
- Bartlett’s Test of Sphericity: Tests whether the correlation matrix is suitable for factor analysis (p < 0.05 indicates suitability).
How to Perform Exploratory Factor Analysis in SPSS
Step 1: Open Your Dataset
For this example, use the following survey data measuring employee job satisfaction:
| ID | Work_Env | Salary | Growth | Benefits | Workload | Job_Security | Manager_Support |
|---|---|---|---|---|---|---|---|
| 1 | 8 | 7 | 6 | 8 | 5 | 7 | 9 |
| 2 | 6 | 6 | 5 | 7 | 4 | 6 | 8 |
| 3 | 9 | 8 | 7 | 9 | 6 | 8 | 9 |
| 4 | 7 | 6 | 5 | 7 | 5 | 6 | 8 |
| 5 | 8 | 7 | 6 | 8 | 5 | 7 | 9 |
- The goal: Identify key job satisfaction factors (e.g., Work Conditions, Compensation, Career Growth).
Step 2: Access the Factor Analysis Tool
- Go to Analyze > Dimension Reduction > Factor.
- A dialog box will appear.
Step 3: Define Variables
- Move all seven job satisfaction variables (
Work_Env,Salary,Growth,Benefits,Workload,Job_Security,Manager_Support) to the Variables box. - Click Descriptives, then check:
- KMO and Bartlett’s Test to assess suitability.
- Anti-Image Correlation to check multicollinearity.
Step 4: Select Extraction Method
- Click Extraction:
- Choose Principal Axis Factoring (better for non-normally distributed data) or Principal Components (default).
- Set Eigenvalues > 1 to retain meaningful factors.
- Check Scree Plot to visualize the number of factors.
- Click Continue.
Step 5: Select Rotation Method
- Click Rotation:
- Choose Varimax (orthogonal rotation for independent factors) or Oblimin (correlated factors).
- Click Continue, then OK.
Interpreting the Output
1. KMO and Bartlett’s Test
- KMO > 0.6: Indicates sufficient sample adequacy for factor analysis.
- Bartlett’s Test p < 0.05: Confirms correlations among variables.
2. Total Variance Explained Table
- Eigenvalues > 1: These components explain significant variance.
- Example: If two factors explain 75% of the variance, these two dimensions capture most of the information.
3. Scree Plot
- Identifies the ideal number of factors (look for the "elbow" where eigenvalues level off).
4. Rotated Factor Matrix
- Shows which variables belong to which factors (loadings above 0.4 are considered strong).
- Example output:
| Variable | Factor 1 (Compensation) | Factor 2 (Work Conditions) |
|---|---|---|
| Salary | 0.85 | 0.20 |
| Benefits | 0.82 | 0.25 |
| Job_Security | 0.78 | 0.30 |
| Work_Env | 0.10 | 0.88 |
| Manager_Support | 0.15 | 0.85 |
| Workload | 0.30 | 0.75 |
| Growth | 0.50 | 0.60 |
Interpretation:
- Factor 1 (Compensation): High loadings for
Salary,Benefits,Job_Security. - Factor 2 (Work Conditions): High loadings for
Work_Env,Manager_Support,Workload.
This suggests job satisfaction has two primary factors: Compensation and Work Environment.
Practice Example: Perform EFA
Use the following dataset of customer experience survey responses:
| ID | Service_Quality | Response_Time | Ease_of_Use | Price_Fairness | Reliability | Trust |
|---|---|---|---|---|---|---|
| 1 | 8 | 7 | 6 | 9 | 8 | 7 |
| 2 | 6 | 6 | 5 | 7 | 7 | 6 |
| 3 | 9 | 8 | 7 | 9 | 9 | 8 |
| 4 | 7 | 6 | 5 | 8 | 7 | 7 |
| 5 | 8 | 7 | 6 | 9 | 8 | 7 |
- Perform EFA to identify key factors in customer experience.
- Use KMO and Bartlett’s Test to check suitability.
- Interpret the Rotated Factor Matrix to define meaningful customer experience dimensions.
Common Mistakes to Avoid
- Using Too Many Factors: Retain factors with eigenvalues >1 or use the scree plot to determine the ideal number.
- Ignoring Rotation: Varimax improves interpretation by making factor loadings clearer.
- Forgetting KMO and Bartlett’s Test: Ensure your data is suitable for factor analysis before proceeding.
Key Takeaways
- Exploratory Factor Analysis (EFA) identifies hidden constructs that explain variable relationships.
- Scree plots and eigenvalues help determine the optimal number of factors.
- Rotated Factor Matrix reveals which variables group together into meaningful factors.
What’s Next?
In Day 35 of your 50-day SPSS learning journey, we’ll explore Confirmatory Factor Analysis (CFA) in SPSS. You’ll learn how to test whether a predefined factor structure fits your data. Stay tuned for more advanced statistical techniques!