Research literacy is the ability to turn a worthwhile question into an investigation that can yield trustworthy evidence without harming the people or environments involved. Students should learn more than the vocabulary of hypotheses and variables. They should understand why a method fits one question but not another, how measurement can mislead, what data cannot establish, and why permission, fairness, privacy and honest reporting are part of research quality.
Ministry of Education V3.0 requires a practical research protocol that can be used across Science, Mathematics, English, Humanities and educational improvement. A class might test a harmless material property, examine an authorised public dataset, evaluate a teaching resource or compare fictional policy options. This reference provides a twelve-stage inquiry method, an ethics screen for projects involving people, guidance on sampling and causal claims, a data dictionary, worked examples, assessment criteria and a correction process.
This is an independent eduKateSG educational design framework, not approval to conduct research involving children, personal health information, vulnerable people, animals or hazardous materials. Schools and researchers must obtain the required institutional and ethics approvals before any relevant real-world study. Students should use teacher-supervised low-risk methods, public datasets or fictional information and never be pressured to participate or disclose private details.
1. A Research Question Determines the Appropriate Method
Different questions need different evidence. ‘How many learners chose an option in a fictional survey?’ is descriptive. ‘Does a particular teaching sequence improve independent algebra performance?’ concerns an intervention and potentially causation. ‘How do people describe a public-service experience?’ may call for qualitative analysis. No single data-collection method answers every question well.
The National Academies Framework for K–12 Science Education sets out scientific practices including asking questions, planning investigations, using models, analysing data, constructing explanations, arguing from evidence and communicating information. These practices help learners understand that scientific knowledge develops through a disciplined relationship between question, method, evidence and revision.
2. Six Questions Before Any Data Collection
| Question | What must be established | Error avoided |
|---|---|---|
| Purpose | What precise claim is the inquiry intended to support? | Collecting data without a valid question |
| Method | Will observation, comparison, experiment or document analysis answer it? | Inferring causation from a descriptive survey |
| Population | Who or what is represented by the material? | Generalising beyond the sample |
| Protection | What approvals, boundaries and safety rules apply? | Creating a risk to participants or environments |
| Measurement | Which variables and units define the result? | Confusing imprecise indicators with capability |
| Return | How will errors, contrary evidence and changes be recorded? | Hiding inconvenient findings |
Good research begins by limiting the claim rather than promising to prove a preferred conclusion. An inquiry may end with ‘The available evidence is insufficient.’ That is often a valuable result when the evidence or method cannot support a stronger statement.
3. The Twelve-Stage Research and Ethics Protocol
Stage 1 — State an answerable question
Write one question with a defined outcome, context and timeframe: ‘Under a fictional scheduling dataset, which arrangement reduces total waiting time?’ is more useful than ‘How can we make the world better?’ Separate the educational curiosity from the preferred policy or outcome.
Stage 2 — Inspect what is already known
Find authoritative explanations and relevant existing evidence. Record original sources, dates and applicability. A classroom should not repeat a widely studied claim as a new discovery, nor trust a generative AI summary without checking its references.
Stage 3 — Choose a method proportionate to the question
Use an observation for a descriptive pattern, a controlled comparison when suitable for testing an explanatory claim, a public-document study for institutional descriptions, or a structured qualitative method for understanding experience. Where a study involves people, appropriate ethical review can be required even if the questions appear harmless.
Stage 4 — Pass the protection and authority gate
Determine whether the activity is permitted, supervised and age-appropriate. Do not assume that a teacher may conduct experiments on pupils’ stress, sleep, health, grades or personal lives simply because the goal is educational. Review risk, consent, privacy, accessibility, safeguarding and the right to decline where applicable.
Stage 5 — Specify what will be measured
Identify the variables, units, categories, data sources and acceptable measurement precision. ‘Student engagement’ is not directly equivalent to app clicks, and ‘health of an ecosystem’ is not equivalent to the number of organisms visible in one photograph. Choose an indicator that is genuinely relevant to the question.
Stage 6 — Define sample and selection
If data come from participants or observations, explain who or what is represented and how selection occurred. A voluntary sample may differ from the broader population. A convenient class sample should not be described as a nationally representative study. Where sampling is not relevant—for example, a mathematical demonstration—do not invent it.
Stage 7 — Identify confounding factors
Ask what else could change the result. A class receiving a new teaching method may also have extra teacher time or different prior knowledge. An outdoor observation may be affected by time of day or weather. Note relevant variables that cannot be controlled and keep the conclusion appropriately modest.
Stage 8 — Predefine interpretation and limitations
Before viewing an outcome, state which comparisons will be made, what would count as a meaningful result and which uncertainty remains. This discourages selecting only favourable numbers afterward. For higher-stakes research, formal protocols, registration and qualified methodological review may be needed.
Stage 9 — Collect or use data safely
Use approved methods and the minimum necessary information. In school, teacher-provided fictional data or legitimate published datasets often supply ample practice. Children should not be required to upload identifiable photographs, interview strangers, handle dangerous equipment or collect personal health information.
Stage 10 — Analyse honestly
Check missing values, measurement errors, consistent units, denominators, comparisons and alternative interpretations. Keep raw observations separate from calculated results. Do not silently delete unusual observations to improve the preferred story; document any justified exclusion rule.
Stage 11 — Report what the evidence supports
Distinguish observation, association, causation, interpretation and policy recommendation. Include limitations and relevant contradictory results. A clear report can be persuasive without pretending that one small investigation establishes a universal law.
Stage 12 — Preserve correction and learning
Record the source, task conditions, method, analysis, interpretation and appropriate review date. If a calculation or assumption later proves wrong, correct it visibly. When the result concerns educational improvement, decide whether a new investigation or safer pilot is needed before wider action.
4. Children Are Participants With Rights, Not Convenient Data
UNICEF Innocenti’s Ethical Research Involving Children compendium places dignity, equitable treatment, protection from harm and informed participation at the centre of research. Research ethics with children must consider their evolving capacities and relevant parental, institutional and legal requirements. Genuine participation is not created by inviting children to answer a compulsory questionnaire whose purpose and consequences they cannot understand.
The compendium’s international charter emphasises informed and ongoing consent and respect for withdrawal, together with the broader ethical duties that apply to children’s research. The practical consent procedures for any real study must be determined by the relevant authority and context. A classroom exercise cannot independently override those procedures.
5. The Child-Participation Ethics Screen
| Ethical test | Question for the adult responsible | Red flag |
|---|---|---|
| Necessity | Must children participate to answer this question? | The same goal is achievable with fictional or public data |
| Benefit | Is there a justifiable purpose and fair burden? | Children bear the risk for an adult’s convenience |
| Permission | Are required approvals and informed participation arrangements in place? | Participation is assumed because pupils are in school |
| Voluntariness | Can the individual decline where applicable without improper penalty? | Coercion, pressure or deceptive explanation |
| Privacy | Is information minimised and secured? | Collection exceeds the stated purpose |
| Risk | Could participation cause physical, social or emotional harm? | Unmitigated exposure, embarrassment or danger |
| Aftercare | Who handles concerns and corrects or withdraws material as appropriate? | No qualified response or redress route |
For most introductory teaching, a fictional dataset is preferable to studying classmates. The educational target is sound method and ethical judgement, not access to real human subjects.
6. Observational Studies and Experiments Are Different
| Design | What it can establish | Important limitation |
|---|---|---|
| Descriptive observation | What was recorded in a defined setting | Does not by itself show why it happened |
| Survey or interview | Reports of views or experiences among participants | Selection and response bias can affect conclusions |
| Comparison without assignment | Differences between defined groups | Pre-existing differences may explain part of the result |
| Controlled experiment | Effect of a manipulation under stated conditions where appropriately designed | Generalisation, implementation and safety still need review |
| Evidence synthesis | What a body of appropriate studies collectively indicates | Quality and comparability of included work vary |
| Simulation or model | What follows from its assumptions and rules | Its outcome may not match real-world behaviour |
An ethical choice of method is as important as a statistically sophisticated one. A learner’s scientific curiosity does not authorise withholding necessary support from classmates or experimenting with their safety, diet, sleep or emotional well-being.
7. Worked Example: A Fictional School Library Study
Illustrative data, not a report of an actual school. A teacher provides a dataset showing how often two types of study spaces were used during different periods. A class asks which space appears more frequently used. The students calculate a rate per available seat and period rather than compare raw visitor counts alone.
Suppose a quiet area has 80 visits across 20 available seats while a collaborative space has 90 visits across 45 seats, under equally defined observation periods. The crude visits-per-seat measures are 4 and 2 respectively. Those figures do not prove that one space is more educationally valuable; repeat visits, length of stay and the purpose of the activities remain unmeasured.
A responsible recommendation might be to gather more appropriate service information before a facilities decision. No pupil needs to follow or identify actual library users to learn this method.
8. Worked Example: Investigating Mathematics Practice
An educator wonders whether a new bridging explanation helps students connect fractions to algebra. A classroom lesson can compare reasoning on appropriate independent tasks before and after instruction, but a single class improvement cannot automatically be attributed exclusively to the new explanation. Prior knowledge, other teaching and task similarity may contribute.
A stronger authorised evaluation could use a valid comparison arrangement and record intervention fidelity, accessibility and independent transfer. The V3.0 Pilot Evaluation reference addresses more consequential programme designs. Ordinary classroom formative assessment and formal human-subject research should not be conflated; consult relevant institutional rules when an activity moves into research.
The Secondary 1 Mathematics small-group example in Clementi illustrates the instructional value of identifying the starting gap before choosing practice. It is not proof of a particular intervention effect.
9. Worked Example: Comparing Public Biodiversity Observations
Two groups receive fictional records of bird sightings from different observation durations. One reports twelve sightings in an hour; the other reports eight sightings in twenty minutes. Students calculate rates per hour under simplified assumptions, but must explain why rate adjustment alone does not remove all detection bias. Species, location, weather and observer expertise can still differ.
This is a low-risk way to teach sampling and data interpretation. The Biodiversity and Ecosystem Services reference extends the subject to habitat and public conservation decisions. For real citizen-science projects, schools need approved observation methods and wildlife protections.
10. The Minimum Research Record
| Field | Minimum content | Purpose |
|---|---|---|
| Question | Precisely stated issue and intended conclusion | Defines the scope |
| Evidence review | Relevant original sources and dates | Prevents unsupported claims |
| Method | Why the chosen design answers the question | Supports validity |
| Ethics and authority | Approvals, safety, data and participation rules | Protects people and institutions |
| Measures | Variables, units and categories | Allows consistent interpretation |
| Sample or dataset | Selection, dates and relevant boundaries | Makes generalisation limits visible |
| Analysis plan | Comparisons and justified exclusions | Reduces selective reporting |
| Results | Observed data and calculations | Separates fact from interpretation |
| Limitations | Bias, uncertainty and missing information | Calibrates confidence |
| Return | Owner, correction and next inquiry | Makes research repairable |
11. A Forty-Five-Minute Classroom Research Design Lab
| Minutes | Activity | Evidence |
|---|---|---|
| 0–7 | Turn a broad question into a specific inquiry | Bounded research question |
| 7–14 | Choose a design and decide which conclusion it could support | Method rationale |
| 14–22 | Screen child protection, privacy and safety | Ethics gate |
| 22–30 | Analyse a fictional dataset with defined units | Transparent calculation and limits |
| 30–38 | Compare plausible explanations and alternative methods | Confounder map |
| 38–45 | Write the qualified conclusion and a repairable record | Short research receipt |
12. Age-Based Progression
| Stage | Appropriate activity | Protection |
|---|---|---|
| Primary 1–3 | Observe non-sensitive classroom materials under teacher supervision | No personal questions or hazardous experiments |
| Primary 4–6 | Compare fictional counts and simple methods | Introduce fair comparison and units |
| Lower secondary | Identify variables, source bias and missing information | Teacher-approved data only |
| Upper secondary | Plan a low-risk investigation and explain ethical limits | Proper approvals for real participants |
| Post-secondary | Evaluate research designs, inference and data responsibilities | Formal institutional ethics rules when relevant |
| Adult learning | Use research evidence in work and civic decisions | Qualified review for consequential studies |
13. The Gold Standard Rubric
| Criterion | Beginning | Proficient | Advanced |
|---|---|---|---|
| Question | Uses an unbounded topic | States an answerable, relevant question | Recognises the decision the result could inform |
| Method | Collects any convenient data | Selects a method matched to the claim | Tests alternative explanations and limits |
| Ethics | Treats available people as research subjects | Recognises permission, safety and privacy duties | Selects a lower-risk design with equivalent educational value |
| Analysis | Presents numbers without definitions | Uses correct units, comparisons and missing-data notes | Evaluates selection bias and robustness |
| Reporting | Declares the preferred result proven | States bounded findings and uncertainty | Documents correction, replication needs and change-of-mind conditions |
14. What a Ministry or School Must Maintain
- Research literacy connected to subject knowledge rather than a separate list of methodology terms.
- Safe, age-appropriate teaching datasets and authorised practical activities.
- Clear differentiation between routine teaching, programme evaluation and formal research involving people.
- Published protection rules for children, data and research environments.
- Teacher training in measurement, sampling, comparison and causal language.
- Source registries and versioned evidence packets for repeated teaching examples.
- Independent marking rubrics that reward truthful uncertainty.
- Procedures for correcting published errors and protecting affected participants.
- Referral to qualified research and ethics professionals when a project exceeds school teaching scope.
15. The Gold Standard
A research-literate learner can pose a clear question, select a method, identify the evidence that would matter, protect the people and environments involved, interpret findings without overclaiming and remain willing to correct their conclusion. They do not need to call every school project an experiment or pretend that gathering more data always improves an answer.
The deeper achievement is a citizen who understands how trustworthy knowledge is made: through appropriate methods, honest limits, respect for participants and repeated return to reality.
Related References and Authoritative Sources
- MOE V3.0 — Complete Reference Library
- Reference 03 — Evidence and Claims Verification
- Reference 09 — Pilots and Responsible Scaling
- Reference 30 — Scientific Uncertainty
- Reference 36 — Decision Laboratories
- Reference 47 — Biodiversity and Ecosystem Services Literacy
- National Academies — Scientific and Engineering Practices
- National Academies — Framework for K–12 Science Education
- UNICEF Innocenti — Ethical Research Involving Children