Did you know that five websites repeating the same impressive fact may still give you only one piece of evidence? Imagine a pupil researching a community garden. Three polished pages confidently say that ’18 plant species were found’, so the pupil treats their agreement as proof. But all three pages may have copied one unverified sentence. Counting web pages is not the same as counting independent observations.
A source chain is the route by which a claim travels from an original observation or document through later summaries, quotations and copies. To test a claim, find where each page obtained it, trace the links back to the original record, check exactly what that record says and seek genuinely independent evidence. Two websites are not independent witnesses merely because they have different URLs or logos.
Why apparent agreement can be misleading
An article may quote a blog; a study guide may summarise the article; a social post may quote the study guide. When the pupil reads all three, the claim appears well supported. Yet the three pieces of writing may all inherit the same transcription error or unsupported assumption. Repetition can amplify confidence without adding new measurements.
This does not mean that a copied claim is necessarily false. It means the number of repeats is not evidence of independent verification. The important question is where the information originated, whether the original is available and whether a later author checked it in a new way.
Researchers call the useful habit of leaving an unfamiliar page and checking other sources lateral reading. A Stanford Graduate School of Education account of research published on 19 April 2022 describes instruction in such source-checking strategies. Our classroom exercise below is an original example of tracing dependence; it does not reproduce that study or claim that its results apply identically to every class.
A fictional school research problem, fully worked
Invented teaching scenario, not a real garden, survey or news item: a class is writing about the imaginary Willow Corner Garden. Webpage A says ’18 plant species were identified’. Webpage B says ‘The garden has 18 species, according to A’. Webpage C repeats B’s sentence without linking to A. An archived field report R, finally located by the students, describes a specific survey of four square sampling plots and lists 12 different plant species recorded in those four plots. No independent new count is provided by B or C.
Step 1: write down one testable claim
Write: ‘An original survey found 18 plant species in Willow Corner Garden’. This is more testable than a vague question such as ‘Is the garden biodiverse?’ Notice that the pupil’s wording includes original survey and garden; both words matter. The field report could support a narrower statement about the sampled plots, not an absolute statement about the entire garden.
Step 2: trace where each page obtained the number
- Page A: displays the number 18 but gives no field-sheet page or table. Status: secondary assertion with uncertain basis.
- Page B: explicitly cites A for the number 18. Status: derivative of A, not a second observation.
- Page C: copies wording from B. Status: derivative of B, which itself depends on A.
- Field report R: provides a date, sampling procedure, four plots and a list of 12 distinct species in those sampled plots. Status: original evidence for that defined survey, subject to evaluating its methods.
The chain is R (possibly relevant but not shown to be A’s source), A → B → C. The student must not invent an arrow R → A merely because the archive report is on the same topic. A could have counted plants elsewhere or misunderstood another source. Unless the connection is documented, mark it ‘not established’.
Step 3: count independent evidence rather than links
There are three webpages stating 18, but they form one visible reporting lineage. They do not represent three fresh surveys. There is one distinct archival report giving a different number under a narrower measurement scope. The class cannot calculate a meaningful ‘three out of four sources agree’ percentage because the sources are not independent votes and they do not even necessarily measure the same population.
Step 4: read the original report’s scope without expanding it
Report R records 12 different species within four sampled plots. From that alone, a pupil may correctly write: ‘The archived survey recorded 12 distinct species in its four plots.’ The pupil may not conclude that the entire garden contained exactly 12 species, or that the claim of 18 for some broader area is necessarily false. Absence from a sample is not proof of absence everywhere. This is where source reading becomes scientific reasoning.
Step 5: write an evidence-calibrated conclusion
A defensible report says: ‘Three online pages repeated a claim of 18 species, but pages B and C relied on earlier pages rather than new counts. A separately located field report listed 12 species within four sampled plots. The two claims differ in evidence and scope; an original record supporting 18 has not yet been verified.’ This is useful, precise and honest even though the research question remains open.
Build a simple provenance table
Instead of opening twenty tabs and hoping to remember who said what, construct a small research table with columns for claim, source URL or archive ID, publication date, quoted evidence, source it cites, whether that citation was opened, survey scope and status. A student’s notebook does not need fancy software. The point is to make dependencies visible.
For our fictional example the entries would be A → unverified origin; B → A; C → B; R → original field report, four plots, twelve listed species. If another team independently surveys eight plots and reports 16 species, that would be a separate observation, but the pupil must still ask whether the plots, dates and identification rules match. Independence alone does not make unlike measurements directly comparable.
The diagnostic: four questions that show what the learner actually understands
- Question A: Two websites repeat the identical sentence and the second links to the first. Is that automatically two independent confirmations? No.
- Question B: A journal article cites a survey. Which should the learner inspect for the original measurement method? The survey itself, when accessible.
- Question C: A survey records 12 tree species in five transects. Can the pupil say the whole forest has only 12? No.
- Question D: The original report is unavailable. Should the pupil invent its sample size from later articles? No; record the source as not yet verified.
- Question E: Two genuine independent counts disagree. Which one is automatically correct? Neither; examine date, location, method and uncertainty.
If A is wrong, the gap concerns evidence independence. If B is wrong, teach primary-versus-secondary evidence. If C is wrong, teach sampling boundaries. If D or E is wrong, teach the difference between an honest unresolved question and a confident unsupported answer.
Common reasons students are misled
- Different website designs: makes copied text appear like a fresh finding.
- Long reference lists: can distract from whether the one relevant statistic actually has an original source.
- Broken links or missing access: tempt a pupil to fill gaps from memory instead of noting a limitation.
- A familiar organisation name: may increase confidence but does not remove the need to inspect the actual claim.
- Changes of wording: a summary may turn ‘found in sampled plots’ into ‘the total in the garden’.
- Publication dates: two 2026 webpages may be copies of the same older field record rather than independent 2026 measurements.
From guided work to genuinely independent research
Level 1: draw a four-node source chain
Give pupils four imaginary source cards: a signed experiment log, a school newsletter citing the log, a local roundup citing the newsletter, and a student essay citing the roundup. Ask them to draw arrows from each retelling to its source. The correct diagram has one observation and three retellings. Then ask what new evidence would be needed to check the first observation.
Level 2: detect a changed meaning
Present the invented sentence ‘Ten insects were observed during a fifteen-minute search on Tuesday’. Then show two paraphrases: ‘Ten insects live in this location’ and ‘Ten insects were observed in one fifteen-minute search’. The second preserves the observational limit; the first converts one timed observation into a population statement. Have pupils identify the extra assumption and repair the sentence.
Level 3: work with a disagreement
A second invented survey reports 14 species in six plots using a later visit. The pupil should not simply average 12 and 14 to obtain 13 ‘true species’. The samples cover different areas and dates; the task is to understand their procedures before attempting a synthesis. Ask the pupil to specify what comparison would make the counts more compatible, such as matching plot areas, observation periods and species-identification rules.
Level 4: use a real, low-stakes information question
With a teacher or parent, choose a publicly documented school-history, museum or science topic. Record one claim and identify three online places repeating it. Follow their cited links to the earliest accessible evidence, noting any dead ends. Summarise in two sentences what has been established and what remains uncertain. Never download private pupil data or contact people merely to force a finding.
A parent-and-teacher routine that avoids the detective-game trap
Parents can ask two questions during homework: ‘Who measured this?’ and ‘Did the next page measure it again, or just repeat the first page?’ These encourage investigation without teaching children that every website is dishonest. The aim is confidence proportional to evidence, not permanent suspicion.
Teachers can mark two different skills separately: a correct trace of the citation chain and a careful statement of what the earliest source supports. A student may find the right first document but still exaggerate its scope. A short oral explanation of the arrow diagram often reveals that error faster than a longer copied paragraph.
Observable progress is a pupil who can, unaided, label a claim as independently observed, derived from another source or unresolved; follow at least one citation to its actual document; identify a shift of meaning; and write a bounded conclusion. Repeat this check on an unfamiliar claim after several days. Speed alone is not the measure of success.
Limits: what a source-chain audit cannot prove
A chain audit may reveal dependence, not truth. An original witness may misremember, a first experiment may use a weak design, and a later independent study may be stronger. Missing citations can arise from broken archives rather than bad faith. Do not accuse a writer of deliberate fabrication merely because an attribution chain is incomplete. Conversely, do not elevate an original source to unquestionable authority. The method and relevant expertise still matter.
Sometimes a professional synthesis accurately combines many primary studies without repeating any single source’s error. This exercise is not a rule that only primary documents are worth reading. It is a rule that the student must know whether apparent agreement comes from independent evidence or a shared ancestor.
Frequently asked questions
Is three agreeing websites stronger evidence than one?
It can be, if they provide independent, relevant evidence. If all three repeat one article, the additional webpages do not provide three independent observations.
What if I cannot reach the original paper?
Report that the original has not been checked and cite the accessible source appropriately under the required citation style. Search a publisher page, library catalogue or legitimate repository. Do not claim to have read a document you have not opened.
Is an older source always better than a newer summary?
No. The older source may be outdated or inaccurate. Trace the provenance to understand the claim, then check whether later research, corrections or more appropriate methods change the interpretation.
Can two sources be independent if one mentions the other?
Possibly. A later report may perform its own measurements and cite a prior study for background. Check its actual methods and data, not merely whether a citation exists.
What should a pupil do when sources disagree?
Compare exactly what each observed, at what time, in which population or location, with what method and limitations. A discrepancy can become a productive research question rather than something to conceal.
What the core aim of education looks like in this task
Learning is not collecting a large number of web links; it is developing the judgment to reconstruct how knowledge was made. The related Core Aim guide to checking a shortened historical quotation gives another source-comparison exercise. The verified library research hub develops responsible evidence use. Browse the eduKate Education Hub for wider transferable inquiry skills. The Clementi small-group mathematics reference models the same progression: find the misconception, show the mechanism, practise gradually and check understanding independently.