Installs into .claude/skills of the current project.
Are you the author of Design Fact Checking Process?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/jeffreytse-design-fact-checking-process)
---
name: design-fact-checking-process
description: Use when establishing or executing a systematic fact-checking process for journalism, fact-checking organizations, or content verification workflows
source: IFCN (International Fact-Checking Network) code of principles; Poynter "MediaWise" fact-checking curriculum; Craig Silverman "Verification Handbook" (2014)
tags: [journalism, fact-checking, verification, misinformation]
verified: true
---
# Design Fact-Checking Process
Build a rigorous, transparent, and reproducible fact-checking process that earns reader trust and meets professional standards.
## Why This Is Best Practice
**Adopted by:** PolitiFact, FactCheck.org, Snopes, Full Fact (UK), Africa Check, and all IFCN-certified fact-checking organizations (90+ globally).
**Impact:** IFCN signatories produce corrections that demonstrably reduce belief in false claims by 15–30% per experimental research (Nyhan & Reifler 2015); standardized processes reduce internal errors by 60%+ vs. ad-hoc verification.
**Why best:** Fact-checking credibility depends entirely on process transparency; without a documented methodology, corrections lack authority.
Sources: IFCN Code of Principles (2016); Silverman "Verification Handbook" (2014); Poynter MediaWise curriculum; Graves "Deciding What's True" (2016).
## Steps
1. **Select the claim** — identify a specific, verifiable, consequential claim. Prioritize: widely shared, made by influential figure, potentially harmful if false. Exclude: opinions, predictions, satire.
2. **Record the claim verbatim** — capture the exact wording, source, date, and context. Paraphrasing introduces distortion before verification begins.
3. **Identify the claim type** — categorize as: statistical/numerical, causal, historical, comparative, or definitional. Each type requires different verification approaches.
4. **Trace the claim to its origin** — follow the claim upstream to its original source. Most viral misinformation is a mutation of an older, sometimes legitimate, claim.
5. **Find authoritative primary sources** — for statistics: government datasets, peer-reviewed studies, official reports. For events: official records, contemporaneous documents, eyewitness accounts with documentation.
6. **Consult subject-matter experts** — identify experts with no stake in the outcome. Document their credentials and any potential conflicts of interest.
7. **Apply the SIFT method** — Stop, Investigate the source, Find better coverage, Trace claims to original context before concluding.
8. **Reach the original claimant** — contact the person or organization that made the claim. Request their evidence. Document all contact attempts and responses.
9. **Apply a transparent rating scale** — use a documented, publicly defined scale (True/False/Misleading/Unverifiable or equivalent). Apply the scale consistently across all checks.
10. **Publish methodology with the verdict** — explain what you checked, what you found, and why you reached your verdict. Link to all primary sources. Publish corrections prominently.
## Rules
- Never rate a claim without contacting the claimant first.
- Publish the full methodology, not just the verdict.
- Issue corrections when new evidence changes the verdict; never silently delete.
- Disclose all funding sources and editorial independence from funders.
## Common Mistakes
- **Checking the wrong claim** — verifying a literal restatement rather than the underlying implication the audience understood.
- **Trusting secondary sources** — news articles reporting on studies are frequently inaccurate; always read the primary source.
- **Missing context** — a technically true claim can be deeply misleading; evaluate the impression it creates, not just its literal accuracy.
- **Inconsistent standards** — applying stricter scrutiny to claims from one political side creates the appearance of bias even when findings are accurate.
## When NOT to Use
- When the claim is clearly an opinion or prediction (not verifiable as true/false).
- When the claim has been definitively and recently fact-checked by a credible IFCN signatory (amplify instead of duplicate).
- When you lack access to the primary sources needed to verify the core claim.