NOETRION

Choose a topic

LESSON 2 OF 7 · AI & information literacy
DIGITAL LITERACY · 5 MIN READ

How to check online information before you share it

A repeatable process for tracing claims, dates, sources, and missing context.

A surprising screenshot gives you a claim, not its provenance. Before sharing it, establish who made the statement, what evidence supports it, and whether the screenshot preserves the original context. This guide gives you a short investigation routine and a worked example you can adapt to technology news, study advice, and everyday claims.

You do not need to become an expert in every subject. You do need to recognize when a claim outruns the available evidence and preserve enough information for another person to repeat your check.

Turn the post into a precise claim

Imagine a synthetic post: “A new app doubles learning speed for every university student.” It sounds specific but hides several questions. Does “speed” mean reading time, quiz score, or time to reach a defined level of mastery? Does “every” mean all students, or only the people participating in one experiment?

Write the claim in a form you can test: “In the cited study, students using the app reached a specified quiz score in half the time of the comparison group.” If the original evidence does not specify a comparison or outcome, mark those elements as missing instead of filling them in yourself.

A claim has several independently checkable parts
Part of the claimQuestion to askWhy it matters
SourceWho produced the evidence?A repost can obscure the origin
PopulationWhich participants were studied?One group does not establish a universal result
OutcomeWhat was measured?Completion speed differs from retained knowledge
ComparisonCompared with what?A benefit needs a meaningful baseline
TimeWhen did the event and publication occur?Old evidence may be presented as new

Read laterally before trusting the page

A professional-looking design and a long “About” page do not establish credibility. Leave the page and investigate its publisher through other sources. The Civic Online Reasoning curriculum calls this lateral reading: opening other sources to examine who is behind an unfamiliar site.

Look for the organization named in the post, its original publication, and independent descriptions of its role. A university press office is a useful source for an announcement, but a press release is not interchangeable with the underlying research paper. A company is authoritative about what it launched, while its claim of superiority still needs an appropriate test.

Follow the evidence chain

Repost or screenshotOriginal announcementUnderlying study or datasetExact supporting result
Original teaching diagram: each link should clarify where the claim came from.

Search distinctive wording in quotation marks. If the post names a study, locate its title and authors. Open the full text or the official record when available, then examine the measured outcome and stated limitations. For a regulation, find the authoritative legal text and the applicable date rather than relying only on commentary.

Keep separate notes for what you can establish and what remains unavailable. An inaccessible paper does not make its findings false, but it limits your ability to verify detailed claims. You can report that you located an abstract while declining to endorse an unsupported numerical interpretation.

Test the learning-app example

Suppose the original announcement describes a 40-person pilot lasting one week. It reports that the app group finished practice questions faster. It does not report delayed recall or include students from every degree program. Our synthetic viral claim has expanded a narrow completion-time result into a universal claim about learning.

A defensible restatement would be: “In this small, short pilot, the app group completed the specified practice task faster. The evidence described here does not establish doubled long-term learning across all university students.” That preserves a potentially interesting result without importing a stronger conclusion.

Ask whether the study randomly assigned groups, whether they had comparable starting knowledge, and whether the same task and scoring rule were used. These are useful investigation questions, not a guarantee that one design feature settles the entire result.

Check dates, denominators, and definitions

An article published today may describe an event from years ago. A chart labeled “50% improvement” might compare two percentages, two raw counts, or a relative change. Record the starting value, ending value, and denominator before repeating the number.

For example, moving from 20 correct answers to 30 is a 50% relative increase in the count. Moving from 20% to 30% correct is a 10-percentage-point rise and a 50% relative increase. Reporting merely “up 50%” leaves the reader guessing which comparison was made.

Seek independent confirmation

Five sites reproducing the same press release are one evidence chain, not five separate investigations. Compare sources that use their own reporting, data, or analysis. When definitions differ, explain the difference instead of treating unequal results as an automatic contradiction.

Google's information about a web page describes tools for learning more about sources. Treat such context as a starting point. It does not replace reading the document that supports the specific statement you plan to share.

Finish with a clear decision

  • Supported: the source directly establishes the stated claim.
  • Partly supported: retain the finding but narrow the wording.
  • Unresolved: say which necessary evidence is missing.
  • Contradicted: explain the conflict with the original source.

Save the source URL, title, relevant section, and date checked. If you shared an error, correct the specific statement and link the better evidence. Good verification makes the uncertainty visible and leaves a path another reader can follow.

Sources and further reading

Source links checked October 13, 2026. Worked examples are original teaching examples; they are not research findings or product benchmarks.

Check your understanding

What best verifies a screenshot of a surprising announcement?

Choose one answer