User Guide: Validity Table Checker

This tool evaluates discriminant validity using fixed mathematical rules. It parses the supplied numbers, compares them directly, and generates the diagnostic report locally.

1. Accepted Table Formats

Paste tables copied from Excel, SmartPLS, SPSS, or another program. Tabs are recommended. Commas, semicolons, Markdown pipes, and clearly separated spaces are also supported. The first row must contain construct names, and each later row must begin with its construct name.

2. Fornell-Larcker Criterion

The square root of AVE on the diagonal must be strictly greater than every absolute correlation involving the same construct.

Sample data included in the app:

Construct	Usability	Effectiveness	Engagement
Usability	0.818	0.695	0.632
Effectiveness	0.695	0.860	0.838
Engagement	0.632	0.838	0.860

All three constructs meet the criterion because each diagonal value is greater than the correlations in its row and column.

3. HTMT Ratio

Below .85: Excellent under the conservative criterion.
.85 to .90: Acceptable under the liberal criterion, but conceptual justification is recommended.
Above .90: Violation of discriminant validity.

Sample HTMT data included in the app:

Construct	Usability	Effectiveness	Engagement
Usability
Effectiveness	0.790
Engagement	0.840	0.880

The .79 and .84 values are classified as Excellent. The .88 value is classified as Acceptable and should be justified theoretically.

4. When a Criterion Fails

Inspect measurement items: Examine outer loadings, cross-loadings, item wording, and indicator correlations. Remove an item only when theory and statistical evidence both support the change.
Reassess conceptual overlap: Determine whether two constructs are defined or measured too similarly. A merger or higher-order construct should be considered only when theoretically defensible.