Did you know that a presentation can have a long reference slide and still contain unsupported claims? A group gives an impressive talk, quotes three websites at the end, then says “most students prefer this” without saying whose students, which survey or what question. More links do not automatically make that sentence reliable. The audience needs a trail from the specific claim to the evidence that can actually support it.
Quick answer: cite the source of a non-obvious fact, borrowed idea, number, quotation, image or disputed claim at the point where the audience encounters it. Explain the method when the group collected its own data. Show calculations as calculations, distinguish inferences from observations, and label invented classroom examples clearly. General knowledge and the group’s own clearly marked opinion may not need a separate citation, but they should never masquerade as independently proven facts.
- Underline every factual, numerical, causal or comparative statement on each slide.
- Classify each as source-based fact, own observation, calculation, inference, opinion or open question.
- Attach a specific reliable source to claims that depend on information outside the group’s own work.
- Make an evidence link visible on the slide, in speaker notes or in an accessible accompanying source list.
- Check what each source actually says; a relevant-looking link is not necessarily support.
- Describe sample size, question wording and collection date for the group’s own surveys.
- Keep paraphrases attributed, not only direct quotations.
- Remove or qualify a claim that cannot be checked.
A 90-second diagnostic for a presentation rehearsal
Read four fictional slide lines aloud. A: “Our group asked 20 classmates this week.” B: “Twelve selected option A on our questionnaire.” C: “All teenagers prefer option A.” D: “This proves option A causes better learning.” Ask which lines could be supported by the group’s own labelled data. A and B could, if the questionnaire and tally exist. C wrongly expands 20 local responses to a population claim. D adds a causal assertion that the poll does not test.
The diagnostic detects two failures often missed by a polished bibliography: changing the population and changing a description into a cause. Students can fix both without inventing another source. The revised talk might say, “In our illustrative class survey of 20 respondents, 12 selected A. This shows their stated choice on this question, not what all teenagers think or which option improves learning.”
Six kinds of statements and what to do with them
1. A published fact or statistic
“The library had 50,000 visits last year” needs the named library, correct year, relevant measure, and a verified publication or dataset. Make sure the figure is visits rather than unique visitors. If the public source uses a different definition, rewrite the slide instead of bending the evidence to fit the headline.
2. A direct quotation
A speaker repeats someone else’s exact words. The speaker or writer must be identifiable and the quoted words checked against the source. Keep quotation marks clear and supply the required citation. A slogan attributed to “experts” but found only on an unattributed image is not a verified quotation.
3. A paraphrase or borrowed interpretation
Changing a sentence’s wording does not make the underlying idea newly invented by the presenter. Paraphrased research still needs attribution. Purdue University’s guide to paraphrasing explains why borrowed information remains source-dependent when expressed in the student’s own words.
4. The group’s own recorded observation
“We counted 18 visitors entering the school exhibition during our ten-minute observation” does not need an external article saying 18. It does need an honest method note: where, when, for how long, what counted as a visitor and whether the observation was permitted. Do not call an illustrative example a real school observation.
5. A calculation from stated information
“15 out of 20 is 75%” is arithmetic. Show the denominator and the operation. If the 15 and 20 came from an external dataset, cite that dataset. The calculation does not gain independent authority merely because it is mathematically correct: it can still start from the wrong population or an outdated number.
6. An inference, prediction or recommendation
“Our group thinks option A might increase participation” should be introduced as a proposal or hypothesis. Readers should see which evidence motivates it and what remains untested. A future claim may require data or a trial before it can reasonably be stated as a result.
Worked example: two surveys, four very different claims
In this deliberately invented teaching example, the same class size is used in two separate surveys. In week 1, 12 of 20 respondents chose option A; in week 2, 15 of 20 respondents chose A. Assume each tally is correct, that the question wording and answer choices match, and that responses were anonymous. We do not assume the same individual students answered twice.
Calculate the percentages openly: 12 ÷ 20 × 100 = 60%; 15 ÷ 20 × 100 = 75%. The difference is 15 percentage points, not 15%. The relative increase in the share selecting A is (75 − 60) ÷ 60 × 100 = 25%. The count of selections rose by three, from 12 to 15, but that alone does not demonstrate that particular children changed their minds.
A defensible slide says, “The share selecting A was 60% in week 1 and 75% in week 2 in these two illustrative 20-person surveys; a difference of 15 percentage points.” Label the table as a fictional classroom example and show the count, denominator and question. An indefensible slide says, “Students are becoming 25% smarter because A works.” Nothing in these invented preference data measures intelligence or establishes cause.
Suppose a presenter instead says, “Published research proves choosing A raises marks.” Now the claim is about an intervention and an outcome; it needs credible original research appropriate to that claim. If the group cannot find evidence, it can explain what study would be needed, or remove the sentence. Citation is not decoration. It constrains what we are entitled to say.
Build a small claim-to-evidence ledger
- Claim: write the exact sentence that will appear on the slide.
- Kind: fact, quote, original observation, calculation, inference, opinion or question.
- Evidence owner: dataset, publication, interview record, group tally or openly declared assumption.
- Locator: page, table, figure, timestamp or precise URL, not just a homepage.
- Check: does the evidence support the wording, population, period, scale and degree of certainty?
- Decision: keep, revise, label illustrative, request better evidence, or remove.
One compact line per substantial claim is enough for most short student presentations. The ledger can live in speaker notes or a shared worksheet. A listener need not be shown the entire internal draft; they should be able to recover the source from a small visible citation or attached reference list.
Do citations make a claim automatically true?
No. A linked page could be wrong, outdated, irrelevant or about a different population. A source might say the survey asked adults in one city while a slide announces a result for all children. A trustworthy presentation checks whether the evidence fits the exact wording and whether another credible source tells a different story. For historical objects and documents, the Library of Congress primary-source teaching guidance suggests moving from observation to questioning, context and corroboration.
A group should also distinguish source credibility from source authority. The organisation named in the browser address does not guarantee that a particular webpage contains the current answer to the question. Use the relevant primary data, original document, official guidance or carefully documented analysis, depending on the claim.
Practice progression: from identifying to defending
Guided — mark the claims
Take a three-line mock slide: “Water freezes at 0°C under standard atmospheric pressure”; “We recorded 0°C at our school table”; “This is the best experiment ever.” Students should recognise a scientific statement with conditions, a school measurement requiring a method record, and a subjective judgement. Ask what could make each line more precise rather than attaching random websites.
Paired — improve one overclaim
Give the line “Every student likes online lessons” beside a fictional tally of 6 supporters among 10 respondents. The revised wording is “In our invented ten-person sample, 6 chose online lessons as their preferred format.” That is 6 ÷ 10 = 60%, with limits clearly stated. The child should explain why 60% of this sample is not “everyone”.
Independent — audit a slide without its speaker
Give a learner an unfamiliar slide with two claims, a graph and a link at the bottom. Ask which figure the link actually supports, whether the title matches the axis labels, and what question remains unanswered. Success means the learner can pinpoint the missing connection, not merely demand “more references”.
Transfer — let another group verify
Exchange source ledgers before rehearsal. One group attempts to trace two claims to their evidence without help. If they cannot identify the exact source table or passage, the presenting group improves its labels and notes. This is a check of communicative clarity as well as academic care.
Teachers, parents and learners: useful next steps
Teachers can ask each student to defend one source link and one own-data calculation during group rehearsal. Parents reviewing slides at home can ask, “How do you know this sentence is true?” without doing the research for the child. Students can practise three answers: “Here is the original source”; “Here is our recorded method”; and “This is our inference, not an established fact.”
Observe progress by how quickly students can locate their own evidence, whether they voluntarily narrow an exaggerated claim, and whether a fresh listener understands the source trail. Do not reward a presentation solely for the number of citations, because a citation flood can make the most important evidence harder to find.
FAQs that prevent unnecessary research work
Does every sentence need a citation?
No. Use citations where they let the audience check borrowed or factual information. Ordinary transitions and clearly identified personal preferences do not necessarily need independent sources. Follow the teacher’s required citation style.
Can we cite our own classroom poll?
Yes, describe it accurately as group-collected data with its date, question, sample size and limitations. Do not fabricate a formal published report, and do not expose participants’ identities.
What if two credible sources disagree?
Check their definitions, dates and populations. A good presentation can state the disagreement and what would resolve it rather than selectively citing only the friendlier result.
Is a source list at the end enough?
A reference list is useful, but the audience may still not know which reference supports which claim. Pair significant claims with a clear in-slide note, numbered source marker or another format prescribed by the teacher.
Continue learning
For precise matching, see how to match citations to references. For the limits of perspective, explore what a source may leave out. Continue through the eduKateSG Education Hub, or use the Clementi Mathematics teaching reference for the same habit of explaining a method before checking it independently.
The core aim is not a beautiful slide with a large bibliography. It is a learner able to say what was observed, what was calculated, what was borrowed, what was inferred, and exactly how someone else can check.