Evidence hierarchy, provenance and limitations

Primary evidence

Structured measurements and results extracted from primary studies are the quantitative authority.

• Reported and normalised values
• Sample and measurement method
• Conditions and source location
• Value origin and estimate quality

Review evidence

Reviews provide taxonomies, consensus summaries, research gaps and secondary benchmarks.

• Always labelled as secondary evidence
• Linked to a primary record only when verified
• Never mixed into primary-value rankings

Publication-safe release

The website contains only allow-listed scientific fields required to interpret the evidence.

• No source documents, abstracts, figures or database files
• No verbatim quotations, local paths or internal audit records
• No extraction notes or internal chapter-planning fields

Attribution, copyright and reuse →

Frozen static architecture

Every dossier is pre-rendered. Search runs against a static index and no live evidence database is deployed.

• One versioned release at a time
• No runtime data writes
• A new export and build are required to update content

What remains unresolved

Property names, units and comparability conditions remain source-preserving until a dedicated quantitative normalisation pass is complete.

• No automated record-value leaderboards
• No averaging across phases, derivatives or sample states
• No comparison across unlike methods, conditions or sample forms

Source verification

Displayed evidence points to a main-text or supporting-information page, section, figure or table where available.

• DOI links lead to publisher records
• Locators are compact, not quoted source text
• Missing locators remain explicit rather than inferred