Trace the source: does it actually support the claim?
Information judgment · Lesson 3: Check sources and context
Estimated time: 20 minutes. Suggested preparation: lessons one and two.
Teaching note: All messages, organizations, dates and figures here are fictional. This source packet is a comparison exercise, not real research or a current notice.
What you will learn
- Trace a report to inspectable source material and record the citation chain.
- Check dates, populations, scope and qualifications.
- Distinguish independent evidence from repeated coverage of one source.
- Record support for individual claims instead of labeling an entire message true or false.
1. Does a link make it credible?
Popular post: Research proves that all students using a learning tool improved their scores by 30%! Three outlets reported it.
Which words most need checking? Do three outlets mean three independent sources of evidence? If you could read one document first, what would you seek?
A link can be genuine while its interpretation is wrong. Ask separately whether the source exists and whether it supports this particular claim.
2. Follow the citation chain backward
A common route is “post → news article → institutional statement → report or data.” Each retelling can add interpretation, omit conditions or alter wording.
Copy the exact claim, then look for links, authors, report titles, publication dates and quoted passages. Without a link, search distinctive phrases, organization names and report titles, recording your search process.
Check the original site's title, publisher, date and corrections. Screenshots, logos and the word “official” do not authenticate a document by themselves. If the original cannot be found, retain an unverified status; failure to find it is not disproof.
Original does not mean correct. Original material lets you examine what was actually stated. Its methods, data and interests still require evaluation.
3. Read three documents side by side
Read this fictional chain from post to report to original project statement. All three documents are supplied here; no external visit is needed.
Document A · Popular post · April 12, 2025: Research proves that all students using a learning tool improved their scores by 30%! Three outlets reported it.
Document B · News excerpt · April 11, 2025: A learning project released trial results. Students completing training and both tests increased their average test score from 50 to 65, a relative increase of 30%. The article cites the project's April 10 statement.
Document C · Original project statement · April 10, 2025: Sixty students volunteered; forty completed training and both tests. These forty students' average score rose from 50 to 65 out of 100. Post-test data from non-completers were not included. There was no comparison group. Results describe score changes among completers only and cannot establish the tool's causal effect. The tool's developer conducted the project.
Between C and A, “completers,” “average” and “no comparison group” disappear. A also turns a group average into a claim about every student.
B's figures agree with C, but B alone does not expose all limitations. C does not prove the tool ineffective either: it describes a change in the stated sample without independently establishing its cause. In a real investigation, inspect methods and data rather than treating the project's own statement as independent confirmation.
4. Check context, not just one sentence
| Check | Compare | Common alteration |
|---|---|---|
| Time | Event, publication and update dates | An old announcement presented as current |
| Population | People, places and tasks covered | Some participants become everyone |
| Measure | The outcome actually measured | Test scores become overall learning ability |
| Conditions | Words such as “may,” “only” and “completers” | Qualifications disappear and certainty increases |
| Context | Surrounding paragraphs, full charts and questions | A quoted opposing view becomes the author's position |
| Version | Revisions, corrections and withdrawals | An outdated conclusion continues circulating |
Keep the exact passage, its location and your check date. This makes later changes to your judgment explainable.
5. Repetition is not independent corroboration
If three outlets cite one project statement, there are three reports but still one originating evidence chain for the trial result. Circular citations do not create new evidence.
Another team collecting its own data may add evidence, but check whether its population, measures and methods address the same question. Independence is not a quality guarantee. Shared sources do not make every article useless: journalists may add interviews or verification.
Record a status for each specific claim:
- Confirmed: Relevant material has been checked sufficiently for a stated scope. For example, you confirmed that the statement says “no comparison group,” not that the tool works.
- Partly supported: Material supports a narrower claim, but not the full assertion.
- Unverified: Necessary material is missing or inaccessible.
- Conflicting: Materials disagree about the same matter; compare versions, populations and evidence quality.
These are review labels, not automatic truth scores. Different claims in one message can have different statuses.
6. Exercise one: how far does the source support the claim?
Using documents A, B and C above, check:
- The project statement reports average scores rising from 50 to 65 among completers.
- Every enrolled student improved by 30%.
- The tool caused the improvement.
For each, cite the relevant passage, assess support and explain your reasoning. Then rewrite the post. Work within this packet; do not treat fictional material as independently verified real data.
Explore the answer and explanation
Explanation
The first claim is confirmed within the packet: C explicitly reports those figures. You confirmed what the statement reports, not the underlying scores independently.
The second claim lacks support. Its scope of all enrollees conflicts with C's analysis of forty completers. An average change also cannot establish the same change for every person. A conflicting label can be justified for the scope mismatch without claiming to know every student's actual outcome.
The third claim is unverified: without a comparison group, this material alone cannot establish causation. C explicitly limits its interpretation. That is not proof that the tool has no effect.
A possible rewrite: “The project reports average test scores rising from 50 to 65 among forty completers. Non-completers were excluded from the post-test analysis, and there was no comparison group, so these results cannot establish the tool's causal effect.”
7. Exercise two: how many evidence chains do you have?
Fictional materials:
Article Alpha cites the project announcement. Beta quotes Alpha. Gamma cites the same announcement. None adds testing. Comment Delta says, “My friend found it useful too,” without records or a measurement method.
Map the citations. How many traceable originating evidence chains support the test-score change? Does Delta independently verify it? What would you seek next?
Explore the answer and explanation
Explanation
Alpha → announcement; Beta → Alpha → announcement; Gamma → announcement. All three lead back to one announcement, not three independent trials.
Delta supplies a different experience claim, but no checkable test-score data or confirmation that it uses the project's measure. It does not independently verify that figure.
Seek methods and data documentation or another team's evaluation using comparable outcomes. If you cannot find independent assessment, record that gap rather than inventing confirmation.
8. Exercise three: an old notice becomes new news
Fictional materials:
Group chat · June 8, 2025: “Every city library branch is closed tomorrow!” Attached is a screenshot without a date.
Original notice · June 8, 2024: “For maintenance, the East Branch will close on June 9, 2024, for one day. Other branches remain open.”
Current situation: The packet contains no opening notice for June 9, 2025.
Identify the date and scope problems and draft a correction for the group. Can you guarantee that every branch will be open tomorrow?
Explore the answer and explanation
Explanation
The screenshot concerns one day last year and only the East Branch. It cannot support a claim about all branches this year.
A correction: “This screenshot comes from the East Branch's 2024 maintenance notice. It is not evidence of a citywide closure this year. Opening arrangements for June 9, 2025, still need confirmation through current notices or direct contact with the library.”
You cannot guarantee that all branches will open. Showing that supplied evidence is inapplicable does not establish the opposite conclusion. Record the screenshot's origin separately from the still-unknown current situation.
9. Review your verification record
Accurately documenting a missing original is useful work. Do not conceal an incomplete check behind a confident answer.
10. A source-checking record
The next unit will examine statistical framing: even accurate figures can mislead when denominators, distributions or comparisons are missing. Continue using the lesson navigation below.
Entries stay in this browser, separately for each language. Export a copy for backup.