Methodology
How CredMeter determines what qualifies as a published claim, how evidence is evaluated, and how the platform handles uncertainty, corrections, and limitations.
What qualifies as a claim
A statement qualifies for publication only when it meets all four criteria:
Attribution: The statement is directly attributable to a specific, named individual. Anonymous or second-hand statements are not accepted.
Source for the statement: The exact or substantially verbatim statement is supported by a reliable primary source — a transcript, recording, official filing, or reporting from a named publication.
Contradicting evidence: At least one primary or official document directly contradicts the statement.
Independence threshold: For high-impact claims, evidence must come from at least two independent domains.
What is excluded
Predictions: Statements about future events cannot be contradicted by current evidence.
Opinions: Value judgments, preferences, and evaluations are not factual claims.
Promises: Commitments to future action are evaluated at a different point in time.
Rhetoric: Hyperbole, analogies, slogans, and clearly figurative language.
Unproven claims: Claims that lack sufficient evidence to confirm or deny are not published.
Accepted source types
Government records: Official statistics, regulatory filings, and agency publications.
Congressional records: Testimony, hearing transcripts, and official legislative records.
Academic datasets: Peer-reviewed publications and institutionally maintained datasets.
Regulatory disclosures: SEC filings, annual reports, and audited financial statements.
Legal records: Court records, FOIA-obtained documents, and official proceedings.
Institutional research: IMF, CBO, Federal Reserve, and comparable bodies.
How contradictions are determined
A contradiction exists when a statement makes a specific, measurable claim — a number, a percentage, a direction of change, a categorical assertion — and an authoritative source provides a materially different measurement or directly opposite finding.
Contradictions are not determined by subjective interpretation. The comparison must be between the same subject, time period, and metric. Statements that are merely incomplete or oversimplified are evaluated separately for a “Misleading” verdict.
From research to publication
- 1.Automated search: Checkable statements are identified and matched against registered sources.
- 2.Evidence scoring: Potential contradictions are scored against the evidence threshold criteria.
- 3.Editorial review: A human editor reviews all claims that meet the threshold before publication.
- 4.Source verification: Evidence sources are independently verified and archived where possible.
- 5.Publication: Claims are published with full source links and a plain-language explanation.
No claim is published automatically. Every published entry has been reviewed by a human editor.
Corrections and challenges
CredMeter accepts correction requests and right-of-reply submissions from individuals who are the subject of a published claim or their authorized representatives.
Corrections are reviewed against the evidence standards. If a correction identifies a genuine factual error, the claim is updated or removed. If the evidence is unchanged, the claim remains published with a note that a challenge was received and reviewed.
To submit a correction: [email protected]
Limitations
CredMeter's coverage is limited by the scope of sources in its register. Absence from the ledger does not imply accuracy. Coverage may be uneven across topics, time periods, and individuals.