claim-confidence-registry

To map manuscript claims to supporting evidence, this Skill locates exact table or cell references in analysis outputs and assigns a four-class confidence taxonomy (robust, tentative, exploratory, unsupported) to flag overclaiming patterns before publication.

Validating Results and Discussion claims for small-N issues involves applying a confidence classification taxonomy to manuscript statements, detecting common overclaiming patterns like small-N subgroup claims, and generating required cautionary language for publication.

Generating a claim matrix links each manuscript claim to its supporting analysis output and assigned confidence class, exporting the complete validated results as a structured CSV or JSON file with required caveats included.

Yes, you can export a manuscript validation report as CSV or JSON. The Skill produces a claim matrix containing mapped evidence cells, assigned confidence classes, and required caveats to ensure transparency and reproducibility.

The confidence classification taxonomy specifically detects cross-validation presented as confirmatory results, applying downgrading rules to overclaims and generating the required cautionary language for those specific validation patterns.

Use claim-to-evidence mapping during manuscript validation when you need to systematically search analysis outputs for exact supporting evidence cells, apply a robust-to-unsupported confidence taxonomy, and prevent null-result language or overclaiming before journal submission.