DataAnalysisHub
Workflow Guide

Copy Loadings from SPSS
Build a CFA in AMOS

Turn your EFA results into a clean CFA diagram effortlessly. Estimate standardized loadings (λ), error variances (θ), and master the Fornell–Larcker check.

What you need before starting

  • SPSS Rotated Component or Pattern Matrix (US1–US6, etc.)
  • AMOS Graphics open with a blank canvas (.amw)
  • Standardized output enabled: Analyze → Output → Standardized estimates
  • Excel template for AVE/CR. Download Here
1

Copy the loading table from SPSS

  1. Find Rotated Component Matrix (orthogonal) or Pattern Matrix (oblique) in Output Viewer.
  2. Right-click inside the table → Copy. Ensure you include the title, header, and item rows.
Tip: The highest loading indicates the component. Cross-loadings should be < .30.
SPSS Matrix Copy
2

Open the Pattern Matrix Importer

In AMOS, navigate to Plugins → Create Diagram from SPSS Pattern Matrix. A text input window will appear.

AMOS Plugin Menu
3

Paste & Create Diagram

Paste your clipboard into the text box and click Create Diagram.

Latent Factors
Created per component
Indicators
Connected to factors
Constraints
One loading fixed to 1.0
Note: If you used Varimax, AMOS may not draw correlations. Manually draw curved double-headed arrows between latent factors for CFA.
4

Rename & Tidy

  • Rename: Right-click factor → Object Properties → Name (e.g., Usability).
  • Clean: Remove unwanted cross-loadings. Ensure simple structure.
Tidy Model
5

Estimate CFA

Calculate estimates and click the "View Output Path Diagram" button. Read the standardized regression weights (λ) and error variances (θ).

Calculate Button
Output Diagram
6

Export to Excel

Calculate AVE, CR & Fornell–Larcker

Workflow
Paste λ in Col B Paste θ in Col D Check Summary Sheet
A PASS flag appears when √AVE > Latent Correlations.

Abbreviations

Estimate: Loading, variance, covariance.
C.R.: Critical Ratio (> 1.96 is sig).
λ (lambda): Standardized loading.
θ (theta): Standardized error variance.

Common Pitfalls

  • Wrong Table: Do not use Structure Matrix. Use Pattern/Rotated.
  • Unstandardized: Always enable "Standardized Estimates".
  • Heywood Case: If λ > 1 or θ < 0, check for multicollinearity.
Fullscreen View