1. Anatomy of the Diagram
Shape Legend
Rectangle
Observed Variable (Item)
Oval
Latent Construct
Small Circle
Error Term (e)
Important Notation
- λi : Standardized loading (from Reg. Weights)
- θi : Error variance (from Variances)
- φjk : Covariance/Correlation (from Correlations)
-
Single-headed arrow (Latent → Item)
Represents the factor loading (λ). This is what you interpret for reliability and validity.
-
Curved double-headed arrow
Represents covariance (φ). In standardized output, this equals the correlation r.
-
Numbers (1)
A "1" on a loading indicates a fixed value used for scaling/identification. The arrow from e to the item is also fixed to 1 by default.
Each latent needs a scale. Either fix one loading to 1.0 (reference indicator) OR fix the latent variance to 1.0. Use only one of these strategies per factor.
2. Building the Model
1. Create Constructs & Items
Use the oval tool for latents (Usability, Effectiveness) and rectangles for indicators (US1...US6). Right-click and use "Object Properties" to rename.
2. Draw Loadings & Errors
Use the single-headed arrow from Latent → Item. Then add error terms (small circle tool) for every item.
3. Scale & Covary
- Pick one indicator per factor: Set Fixed Value = 1.
- Draw curved double-headed arrows between all latents.
4. Estimate
Go to Analyze → Analysis Properties → Output and tick Standardized estimates.
3. Interpretation
Loadings (Latent → Item)
The number is the standardized loading λ.
Error & Variance
Small circles: Path fixed to 1. Error variance θ is estimated.
Double-arrows: Show Covariance (φ). In standardized output, these are correlations (r).
Common Operations
Quality Checks
Do: Keep a simple structure. Ensure each item has an error term. Scale each factor properly.
- Heywood cases (θ < 0, λ > 1)
- Forgetting factor covariances
- Mismatched item numbering