User Guide: AMOS Fit Index Interpreter

Welcome to the AMOS Fit Index Interpreter. In Chapter 7, after drawing your Confirmatory Factor Analysis (CFA) model in AMOS, you must prove that your theoretical model "fits" the actual data you collected. This tool helps you read the complex AMOS output.

1. What does "Model Fit" mean?

Model fit tests how well your proposed theoretical structure (e.g., "These 5 items measure Motivation") matches the reality of your respondents' answers. If the fit is good, your survey is measuring what you think it is measuring.

2. Model Fit "Cheat Sheet"

There are many indices in AMOS, but academic literature (e.g., Hu & Bentler, 1999; Hair et al., 2010) typically requires reporting these four:

CMIN/DF (χ²/df): Ratio of Chi-square to degrees of freedom. Target: < 3.0 (Good) or < 5.0 (Acceptable).
CFI (Comparative Fit Index): Target: > 0.95 (Good) or > 0.90 (Acceptable).
TLI (Tucker-Lewis Index): Target: > 0.95 (Good) or > 0.90 (Acceptable).
RMSEA (Root Mean Square Error of Approximation): Target: < 0.06 (Good) or < 0.08 (Acceptable).

3. The Problem with Chi-Square (CMIN)

You might notice that the p-value for your CMIN is less than .05. In many statistical tests, a significant p-value is good. In model fit, it is bad. A significant p-value means your model is significantly different from a perfect fit.

However, the Chi-Square test is highly sensitive to sample size. If you have over 200 participants, it will almost always be significant (p < .05). That is why researchers ignore the raw CMIN p-value and instead look at CMIN/df.

4. Tracking Model Iterations

When running a CFA, your very first attempt rarely produces a perfect fit. You will likely need to test and adjust your model several times. The Model Iteration dropdown in this app helps you track these stages and provides the correct context for your APA report:

Initial Measurement Model: This is your first run in AMOS before making any changes. It often returns a "Poor Fit." You report this to show your starting point.
Modified Model 1 & 2: These are the intermediate stages where you adjust your model to improve the fit indices. Common modifications include deleting items with standardized regression weights (loadings) below 0.50, or drawing covariances between error terms based on AMOS Modification Indices (MI).
Final Model: This is the cleaned-up model that successfully achieves "Acceptable" or "Good" fit thresholds. This is the definitive model you will report and rely on for the rest of your thesis.

Note: By selecting the correct iteration from the dropdown, the AI will appropriately adapt the phrasing in your generated APA 7 write-up.

5. Where to find the data in AMOS

Run your AMOS analysis.
Click on View Text (the little page icon).
In the left menu, expand Estimates, then click on Model Fit.
Highlight all the text in the main right-hand window, copy it, and paste it directly into this tool.