Methodology

#How topics are selected

Topics are chosen where Weavidence has something genuinely useful to say — either because a methodological question is common and underserved by clear public writing, or because Weavidence's own product engineering illustrates a general principle worth explaining on its own terms, not just as a feature description.

#How questions are scoped

Each article is built around one answerable question, stated explicitly in its front matter (question) and answered directly near the top of the piece (answerBrief) before the full reasoning is laid out. A question that turns out to require several genuinely separate answers is split into separate articles rather than answered partially in one.

#How sources are identified

For methodological and public-health claims, sourcing follows a preference order: official institutional guidance first, then current reporting and methodological standards, then original research, then high-quality systematic reviews, then other authoritative methodological references. For Product Notes, the primary source is Weavidence's own documented architecture and verified implementation — internal documentation and, where relevant, direct evidence from the codebase itself (for example, a passing automated test that proves a described behavior actually occurs).

#How claims are checked

A claim is only published with a citation that actually supports it — not a source that is merely topically related. Every inline citation resolves to a specific entry in the article's reference list, and every listed reference is actually cited somewhere in the body; an article with a reference that supports nothing in the text, or a claim with nothing behind it, fails validation and cannot be published (see scripts/verify-insights.mjs in the Weavidence Landing repository for the automated check).

#When current sourcing cannot be verified

Some environments this platform is built and maintained in have no live network access to confirm that a source's content, URL, or currency have not changed since an author's information was last current. When that is the case, an article whose claims depend on external verification is held in review status rather than published with an unverifiable citation. This is a deliberate, disclosed limitation, not an editorial gap — see the initial Weavidence Insights launch report for two draft articles held in exactly this state.

#How uncertainty is represented

Articles state limitations and uncertainty explicitly, in their own visible section, rather than burying qualifications in a footnote or omitting them. Where evidence is mixed or a claim has real methodological debate behind it, the article says so rather than presenting a contested point as settled.

#How Product Notes differ from evidence summaries

A Product Note describes what Weavidence built and why, grounded in Weavidence's own documentation and implementation — it is a design rationale, not an independent evaluation of Weavidence, and it is labeled as such. An Evidence Brief synthesizes external evidence on a public-health or methodological question and is held to the full external-sourcing standard above. The two are never presented interchangeably, and a Product Note does not borrow the evidentiary weight of an Evidence Brief's content type.

#How a review level is assigned

Editorial review applies by default. Methodological review is assigned to any article that makes a claim about study design, statistical reasoning, or analytical methodology. Clinical review is assigned to any article that touches, even indirectly, a topic that could inform an individual clinical decision — and such an article is not published without an identified reviewer recorded in its front matter (reviewerIds); the automated validation gate enforces this and rejects publication of clinical-level content with no reviewer on record.