A fact check is the process of verifying whether a factual claim is accurate by examining reliable evidence, original sources, and verifiable data. The goal is not simply to decide whether something is "true" or "false," but to explain what the evidence shows, how it was evaluated, and why a particular conclusion was reached.
In simple terms:
A fact check is an evidence-based investigation of a factual claim.
It answers questions like:
- Is the claim accurate?
- What evidence supports or contradicts it?
- Are important details missing or taken out of context?
- How confident can we be in the conclusion?
What is a "fact"?
A fact is something that can be objectively verified.
Examples:
✅ Fact:
- "Dhaka is the capital of Bangladesh."
- "Water boils at approximately 100°C at sea level."
- "The 2024 Summer Olympics were held in Paris."
These can be confirmed using reliable sources.
What is a "claim"?
A claim is a statement that asserts something to be true.
Examples:
- "The government built 500 new schools."
- "This medicine cures diabetes."
- "The photo was taken yesterday."
- "Crime has doubled."
Not every claim is true, and many require verification.
Fact Checking = Claim + Evidence
A fact check always starts with a claim.
Claim
↓
Identify what can be verified
↓
Collect evidence
↓
Evaluate source reliability
↓
Compare claim with evidence
↓
Reach a conclusion
What can be fact-checked?
Many types of factual claims, including:
- Political statements
- Government policies
- Election claims
- Health and medical information
- Scientific assertions
- Historical claims
- Economic statistics
- Viral social media posts
- Images and videos
- Quotes attributed to public figures
- Maps, locations, and events
What cannot be fact-checked?
Some statements are not objectively verifiable.
Opinions
"This is the best movie ever."
This is subjective.
Preferences
"Chocolate ice cream tastes better than vanilla."
Personal preference.
Predictions
"The economy will grow 10% next year."
This cannot be verified at the time it is made, though later evidence may show whether the prediction was accurate.
Beliefs
"This painting is beautiful."
A matter of personal judgment rather than objective fact.
Fact Checking vs. News Reporting
| News Reporting | Fact Checking |
|---|---|
| Reports what happened | Verifies whether a specific claim is accurate |
| May include multiple viewpoints | Evaluates evidence against the claim |
| Focuses on events | Focuses on accuracy |
| Can be immediate | Often requires detailed research |
Fact Checking vs. Research
Research seeks to generate or expand knowledge.
Fact checking seeks to verify an existing claim.
Example:
Research:
"How does climate change affect agriculture?"
Fact check:
"Did Bangladesh lose 30% of its farmland due to climate change?"
Fact Checking vs. Debunking
A fact check is broader than a debunk.
- Fact check: May conclude a claim is true, partly true, misleading, unsupported, or false.
- Debunk: Specifically aims to show why a false or misleading claim is incorrect.
Core Principles of Fact Checking
A high-quality fact check should be:
- Evidence-based – conclusions rely on verifiable evidence.
- Transparent – sources and reasoning are clearly documented.
- Independent – conclusions are not influenced by political or commercial interests.
- Fair – considers evidence that both supports and contradicts the claim.
- Correctable – updates or corrections are made if new evidence emerges.
A Typical Fact Check
Claim
"Country X has the highest literacy rate in South Asia."
Verification
- Check official statistics.
- Compare data from international organizations.
- Confirm the reporting year and methodology.
Finding
The available data show another country has a higher literacy rate.
Verdict
False
Explanation
The claim is incorrect because the latest comparable statistics rank another country above Country X.
Fact Checking Methodologies
There isn't a single universal fact-checking methodology. Different organizations use different frameworks depending on whether they verify a political speech, an image, a viral social media post, or a historical claim.
It's useful to understand both the industry-standard methodologies and how they can translate into a structured database and workflow.
1. Claim-Based Fact Checking (Most Common)
This is the methodology used by most professional fact-checking organizations.
Workflow
Claim
↓
Identify the exact claim
↓
Find original source
↓
Collect evidence
↓
Compare claim vs evidence
↓
Reach verdict
↓
Publish reasoning
Example
Claim:
Bangladesh has the highest literacy rate in South Asia.
Process
- Define exactly what "highest" means.
- Check UNESCO
- Check World Bank
- Check Bangladesh Bureau of Statistics
- Compare latest data
- Publish verdict
Result
Rating: False
Reason:
Sri Lanka has a higher literacy rate than Bangladesh.
2. Evidence-Based Methodology
Instead of beginning with an opinion, start with evidence.
Question
↓
Primary Sources
↓
Secondary Sources
↓
Expert Opinions
↓
Cross Verification
↓
Conclusion
Evidence priority
- Official documents
- Government statistics
- Peer-reviewed research
- Court records
- Original video/photo
- News reports
- Social media
Higher evidence always outweighs lower evidence.
3. Source Reliability Methodology
Every source receives a reliability score.
Example
| Source | Reliability |
|---|---|
| Constitution | 100 |
| Supreme Court | 98 |
| Government Gazette | 95 |
| Election Commission | 95 |
| WHO | 94 |
| World Bank | 93 |
| Reuters | 90 |
| AP | 90 |
| Local newspaper | 75 |
| Blog | 45 |
| Facebook Post | 20 |
| Anonymous Tweet | 5 |
Instead of asking
Is this claim true?
You ask
How trustworthy is the evidence?
4. Triangulation Method
Professional investigators often require at least three independent sources.
Claim
↓
Source A
↓
Source B
↓
Source C
↓
All agree?
↓
Verdict
Example
Election result
Check
- Election Commission
- Reuters
- AP
If all agree
High confidence.
5. Reverse Verification
Especially useful for viral content.
Instead of starting from the claim
Start from the media.
Example
Image
↓
Reverse Image Search
↓
Original upload
↓
Metadata
↓
Location
↓
Date
↓
Conclusion
Used heavily for
- Fake photos
- War misinformation
- AI images
- Old images shared as new
6. Timeline Verification
Useful when dates matter.
Example
Claim
Event happened yesterday.
Create timeline
Police Report
↓
News
↓
CCTV
↓
Weather
↓
Social Media
↓
Witnesses
Many fake stories collapse because the timeline doesn't align.
7. Geolocation Verification
Used by OSINT investigators.
Example
Video claims
Recorded in Dhaka
Verify
- Buildings
- Road signs
- Language
- Google Maps
- Shadows
- Weather
- Terrain
Then determine
Actual location.
8. Metadata Verification
Digital files contain hidden information.
Check
- EXIF
- GPS
- Camera model
- Edit history
- Upload date
Sometimes metadata is removed or altered, so corroboration is still necessary.
9. Context Verification
Many claims are technically true but misleading.
Example
Politician says
Crime increased 50%.
Questions
Compared to what?
One month?
One year?
One city?
Whole country?
Without context
Rating
Misleading.
10. Statistical Verification
Used for numerical claims.
Example
Inflation doubled.
Need
- CPI
- Government statistics
- IMF
- World Bank
Then calculate independently.
11. Scientific Fact Checking
Hierarchy
Peer-reviewed research
↓
Meta Analysis
↓
Systematic Review
↓
RCT
↓
Observational Study
↓
Expert Opinion
↓
Personal Experience
The highest quality evidence generally carries the most weight.
12. Legal Verification
Check
- Constitution
- Laws
- Rules
- Gazette
- Court Judgments
Example
Claim
This law was repealed.
Verify from
Official Gazette.
13. Historical Verification
Use
- Archives
- Books
- Newspapers
- Museum records
- Government documents
- Academic papers
Avoid relying solely on secondary summaries.
14. Primary Source First
One of the strongest methodologies.
Always ask
Where did this information originate?
Example
News article
↓
Press Release
↓
Government Document
↓
Actual Data
Always try to reach the original source rather than relying on repeated reporting.
15. Open Source Intelligence (OSINT)
OSINT investigators combine many techniques.
Typical workflow
Claim
↓
Image
↓
Video
↓
Satellite
↓
Maps
↓
Weather
↓
Flight data
↓
Ship data
↓
Public records
↓
Social media
↓
Verdict
16. Claim Decomposition
Large claims become smaller claims.
Example
The government built 200 hospitals and reduced infant mortality by 50%.
Break into
- Did government build hospitals?
- Was it 200?
- During what period?
- Did mortality reduce?
- Was reduction 50%?
- Is there causation?
Each subclaim gets verified separately before assessing the overall statement.
17. Confidence-Based Methodology
Instead of only True/False.
High Confidence
↓
Medium Confidence
↓
Low Confidence
↓
Unknown
↓
Unverifiable
This better reflects uncertainty when evidence is incomplete.
18. Fact vs Opinion Separation
Always separate
Fact
Dhaka is the capital of Bangladesh.
Opinion
Dhaka is the best city.
Predictions
Bangladesh will become developed by 2040.
Only factual claims can be verified directly.
19. Claim Lifecycle Methodology
Useful for a fact-checking database.
Claim Submitted
↓
Under Review
↓
Research
↓
Evidence Collected
↓
Peer Review
↓
Published
↓
Updated
↓
Archived
This preserves the history of a claim as new evidence becomes available.
20. Editorial Review Methodology
Many organizations require two reviewers.
Researcher
↓
Senior Researcher
↓
Editor
↓
Legal Review (if needed)
↓
Publication
This reduces bias and catches errors before publication.
Common Verdict Systems
Different organizations use different labels, but a comprehensive system could include:
| Verdict | Meaning |
|---|---|
| True | Fully supported by reliable evidence |
| Mostly True | Minor inaccuracies, but core claim is correct |
| Partly True | Mix of accurate and inaccurate elements |
| Misleading | Uses facts out of context or in a deceptive way |
| Unsupported | Insufficient reliable evidence |
| False | Contradicted by reliable evidence |
| Fabricated | Entirely invented or forged |
| Satire | Not intended as factual reporting |
| Opinion | Subjective statement, not fact-checkable |
| Unverifiable | Cannot be confirmed or refuted with available evidence |
| Outdated | Was accurate at one time but no longer reflects current facts |
The hybrid methodology is well suited for a rapid project initiation
Given your goal of building an information repository that can power a fact-checking platform, a hybrid workflow would be particularly effective:
- Claim intake and classification (identify and categorize the claim).
- Claim decomposition (break complex claims into verifiable subclaims).
- Evidence collection with a clear hierarchy (primary sources first).
- Source reliability scoring for every piece of evidence.
- Cross-verification using at least two or three independent sources where possible.
- Context analysis (dates, geography, legal and historical context).
- Confidence scoring alongside a verdict, to communicate certainty.
- Editorial review before publication.
- Versioning and updates, so fact checks can evolve as new evidence emerges.
This hybrid approach is scalable, transparent, and maps naturally to a structured database where claims, evidence, sources, people, organizations, places, and verdicts are all interconnected.