Back to the episode map

Research Note

Claim Source Comparison Record

Source comparison begins with one claim that can be proved, narrowed, or rejected. It does not begin by collecting outlets that appear to represent opposing teams.

Aug 4, 20262 min readBy Dalton Anderson

Claim Source Comparison Record

Core method

Source comparison begins with one claim that can be proved, narrowed, or rejected. It does not begin by collecting outlets that appear to represent opposing teams.

The Digital Inquiry Group research summary Lateral Reading and the Nature of Expertise reports that professional fact checkers left unfamiliar sites, opened new tabs, and investigated the organization and claim elsewhere. The method is useful because a polished page can only tell the reader what its publisher chose to present.

Evidence roles

RoleQuestion
Claim originWho first made the specific statement, and when?
Primary recordWhat document, dataset, filing, recording, code, or direct observation bears on it?
Independent scrutinyWho examined the evidence without simply repeating the origin?
Relevant expertiseDoes the source have methods and domain knowledge suited to this claim?
Disconfirming evidenceWhat credible evidence would narrow or contradict the claim?
Corrections and versionHas the record changed, been corrected, or gone stale?
Incentive and accessWhat relationships, funding, access, or commercial interests matter?

Source diversity is not a vote. Ten articles that all cite one press release are one evidence path. A falsehood does not become balanced because it is placed next to a supported finding. Disagreement is useful when it exposes different data, definitions, methods, or uncertainty.

Conclusion language

End with the narrowest statement the evidence supports. Use "the filing reports," "the experiment found in this setting," or "the available record does not establish" instead of broad certainty. Record the date and the evidence that would change the conclusion.

Worked-example boundary

The public guide may use the claim "personalized feeds cause political polarization." It should separate evidence that feeds personalize exposure from evidence that an intervention changed attitudes. The 2023 Nature Facebook experiment is suitable precisely because its exposure and attitude results differ.

Sources

Follow the evidence.

  1. support.google.com: 14328491support.google.com
  2. nature.com: s41586 023 06297 wnature.com
  3. solomonmg.github.io: bakshy 2015 exposuresolomonmg.github.io
  4. ftc.gov: federal trade commission announces updated advertising guides combat deceptive reviews endorsementsftc.gov
  5. FTC Disclosures 101ftc.gov
  6. support.google.com: 16533387support.google.com
  7. spec.c2pa.org: Explainerspec.c2pa.org
  8. youtu.be: GhR9NuCQwlMyoutu.be
  9. FTC Endorsement Guides questions and answersftc.gov
  10. support.google.com: 2801947support.google.com
  11. cor.inquirygroup.org: lateral reading and the nature of expertisecor.inquirygroup.org
  12. support.google.com: 6342839support.google.com
  13. newsroom.tiktok.com: how tiktok recommends videos for younewsroom.tiktok.com
  14. open.spotify.com: 6pC7h1LY88I1LnfvH6h4dvopen.spotify.com
  15. daltonanderson.ghost.io: ai influencers and the dangers of a filtered realitydaltonanderson.ghost.io
Claim Source Comparison Record