Claims, Sources and Uncertainty
What this concept is
A claim is something that could be supported or challenged. Evidence is the observation or argument offered for it. A source tells you where the claim or evidence came from. Lateral reading means leaving a page to investigate its origin and relevant independent context, rather than judging credibility by its appearance alone.
Why it matters here
An engineer must distinguish a tutorial example, a vendor promise, a measurement and a general conclusion. This habit also protects your study plan from unsupported productivity claims.
Concrete example
An invented advertisement says: “Our learners doubled their success rate.” The data says 4 of 20 volunteers passed before a workshop and 8 of 20 passed afterward. The rate went from 20% to 40%: an increase of 20 percentage points and 100% relative to the original rate. Both statements describe the same numbers.
That does not prove the workshop caused the change. Were these the same learners? Were the tests comparable? Who dropped out? Was there a comparison group? A cautious rewrite is: “In this small reported sample, the pass rate increased; the description does not isolate the workshop's effect.” These numbers are an exercise, not research findings.
Common confusion / misconception
Two articles repeating one press release are not two independent observations. A primary source can still have weak methods or a conflict of interest. A later date does not automatically make a source more reliable. “Insufficient evidence” does not mean the opposite claim is true.
How to use it
Write the exact claim and scope. Find who made it, the original evidence and relevant date/version. Check an independent source for context or disagreement. Inspect sample, denominator, comparison and missing outcomes. Label each statement as observed, inferred or recommended. State what additional evidence would change your conclusion.
If you cannot access the source, mark it unverified. Do not cite an AI-generated title as if you read the paper. The existing responsible-AI strand owns the disclosure log; link to that entry instead of creating another one.
Check yourself
Two websites cite the same survey. Have you replicated the result? No. You have found repeated reporting of one source. Seek different evidence or acknowledge the limitation.
Mini drill or application
Audit one real learning claim from an article or video. Follow its source trail, calculate any simple percentage claim, and write a qualified conclusion. If the original evidence is absent, record the unsuccessful search and narrow the conclusion.
Read this only if stuck
Brodsky and colleagues (2021) studied lateral-reading instruction in college civics sections. Their results motivate practice here; transfer to engineering research must be demonstrated, not assumed.