We select claims that are influential, widely shared, consequential, or useful for teaching how evidence works. We define the claim as precisely as possible before assessing it. A vague or changing claim may need clarification before a verdict is useful.
We look for primary sources: legislation and judgments for legal claims, datasets and research papers for scientific claims, and original records for historical claims. We also consult careful secondary analysis where it adds context. We check whether a source actually supports the statement attributed to it and note limits, jurisdiction, date, and uncertainty.
Verdicts are editorial judgments, never generated from keywords. FALSE means the evidence contradicts the central claim; MISLEADING means true elements create a materially false impression; UNSUPPORTED means adequate evidence has not been supplied; PARTLY TRUE means significant parts are accurate and others are not; CONTEXT NEEDED signals missing context; UNVERIFIABLE means available evidence cannot currently settle the question. An article may remain unlabelled while it is in review.
We compare conflicting evidence, explain the reasoning, and link readers to source records. When new evidence changes an answer, we update the article and publish a correction note. Readers can report errors through the article form.