eduKateSG · Why Science?
Keep the circumstances beside the result, so another reader can understand the experiment
Choose the closest question, or read the guide from the beginning. Each chapter keeps its explanation, example and next step visible.
A result without its circumstances is like a photograph with the edges cropped away. You may see something interesting, yet miss the information that explains it. The Experimental Condition Ontology, or XCO, helps give those circumstances shared names in physiological and morphological measurement records. Why learn Science? Because knowing what happened is stronger when we also know under which conditions it happened.
The OBO Foundry describes XCO as covering conditions under which measurements are made in clinical work and studies involving humans or model organisms. Its official repository identifies the Rat Genome Database as maintainer. This guide owns context around measurement, rather than generic experimental-factor vocabulary or environmental classification. It links those neighbouring owners when they help explain a different part of the record.
For Primary Science, PSLE Science and Secondary Science learners, the starting idea is fair comparison. For advanced readers, it becomes a condition matrix with values, time windows and provenance. Examples here are fictional, non-clinical teaching cases. Suggested designs are not instructions for human or animal experiments, medical advice, official XCO requirements or promises of grades, admission or career outcomes.
Follow the connected evidence: Experimental Factor Ontology; Environment Ontology; Ontology of Microbial Phenotypes. For education and school-choice guidance, use the Education Hub; for possible work directions, continue to Careers by Subject Capability. These routes are separate from the specialist topic here.
| Layer | Question | Record | Boundary |
|---|---|---|---|
| Condition type | Which circumstance is described? | XCO class, definition and release | Not a measured outcome |
| Condition setting | What value or level applied? | Value, unit, material and status | Not fully described by a class |
| Time structure | When and for how long? | Start, duration, sequence and measurement window | Not a timeless label |
| Design role | Why is this condition in the study? | Treatment, control, block or recorded context | Not automatic causation |
Inside this guide
Chapters 1–10 · Meaning and evidence
- 1. Did You Know? Conditions can be the missing explanation
- 2. Separate condition, method and outcome
- 3. Begin with the study question
- 4. Map the type, then record the setting
- 5. Planned conditions and realised conditions can differ
- 6. Time turns a label into an exposure history
- 7. Record material identity without inventing completeness
- 8. Controls are defined by what they control
- 9. Batches and sites are contextual groups
- 10. Randomisation, blocking and recording do different jobs
Chapters 11–20 · Comparison and learning
- 11. Do not convert context into cause
- 12. A shared condition class is a search aid
- 13. Worked case: one label, two durations
- 14. Worked case: a misleading zero
- 15. Worked case: matching type, mismatched phase
- 16. Version context meaning as well as data
- 17. Teach context through safe fair comparisons
- 18. Give parents a manageable conversation
- 19. Connect context literacy to education and work
- 20. Finish with an accountable context bundle
CHAPTER 1 OF 20 · Understand context
1. Did You Know? Conditions can be the missing explanation
The official XCO description centres the conditions under which physiological and morphological measurements are made. That makes context a first-class part of the record rather than an afterthought. However, a condition term alone does not state its numeric setting, duration, implementation or relationship to the measurement. Those details still need explicit fields. Source: the OBO Foundry XCO overview.
Picture two fictional readings of a material's length: one made immediately after moving it into a room and another after a defined waiting period. The result may depend on the object's state, not just the ruler. Without the waiting information, a reader cannot tell whether the observations concern comparable circumstances.
The useful habit is to write a short condition sentence alongside every result. Ask what a second team would need to know to reproduce the context. If the answer lives only in the first observer's memory, the record is unfinished. Shared vocabulary helps organise that sentence, while values and timing make it operational.
CHAPTER 2 OF 20 · Understand context
2. Separate condition, method and outcome
Conditions describe circumstances; methods describe measuring processes; outcomes describe observations. These dimensions can interact, but they should not be merged into one vague field. A measured temperature may be an outcome in one investigation and a recorded condition in another. Its role comes from the study question, not from the word alone.
For a fictional paper experiment, room humidity is context, a ruler-based process is method and strip length is the outcome. If the question changes to monitoring room humidity, the humidity value becomes the outcome. Clear record design follows the scientific purpose instead of assuming every variable has a permanent role.
Recommended fields separate condition type, setting, time and design role. Add the method and outcome through their own references. This allows a reader to ask whether disagreement came from changed circumstances or changed measurement. It also gives students a practical way to organise fair-test answers: what changed, what was measured and what relevant circumstances were kept comparable?
CHAPTER 3 OF 20 · Understand context
3. Begin with the study question
You cannot decide which conditions matter without knowing the question. A study comparing two settings needs a different record from a study describing variation over time. NIST's design guidance says design selection depends on experimental objectives and the number of factors. That principle is broader than XCO, but it explains why context should be selected deliberately. Source: NIST experimental-design selection.
Write the question in one sentence, then identify the intended comparison and observation unit. List conditions that could change the interpretation. Do not collect every imaginable field simply because a database can store it. Prioritise the circumstances that make the question reproducible and the comparison meaningful.
For students, this is a calm way to start an unfamiliar experiment problem. Before naming variables, ask what the investigators are trying to distinguish. A good answer can be short: “They want to compare the same endpoint under two stated settings.” Once that purpose is clear, method, condition and result fields can be arranged around it rather than memorised as disconnected labels.
CHAPTER 4 OF 20 · Understand context
4. Map the type, then record the setting
A controlled term tells readers what kind of condition is being described. It does not necessarily provide the value applied in one experiment. Keep the source label, selected XCO identifier, definition and release, then store the setting separately. A number without a condition type is ambiguous; a condition type without a setting can be too broad.
Use fictional examples rather than unverified ontology identifiers: “temperature setting” with a recorded value and unit, or “waiting condition” with a stated interval. These labels illustrate record structure and are not asserted to be exact XCO classes. Real annotation requires consulting current official definitions and selecting a supported term.
Recommended mapping review asks whether the source circumstance fits the term, whether important qualifiers remain outside it and whether the mapping is exact or broader. Preserve uncertainty. If a source says only “standard conditions”, do not infer a numeric temperature, lighting cycle or preparation regime. The honest result is a documented gap, with a request for the missing detail.
CHAPTER 5 OF 20 · Understand context
5. Planned conditions and realised conditions can differ
A protocol can specify a target setting while the actual environment varies. Record both when the difference matters. A planned value describes intention; a measured value describes what was observed. Treating the two as identical can hide deviations and make a dataset appear more controlled than it was.
In a fictional classroom case, a room is intended to remain at one temperature but readings vary during the activity. The condition record should show the target, observed range, measurement times and any relevant interruption. It should not rewrite all observations as the planned value to simplify the table.
Recommended review distinguishes acceptable variation, uncertain monitoring and protocol deviation. State the criteria used and retain the original record. A deviation need not make all evidence useless; it may narrow the comparison or require another analysis. What matters is that readers can see it. Science learning becomes more realistic when students understand the difference between a neat plan and an accountable record of what happened.
CHAPTER 6 OF 20 · Describe conditions precisely
6. Time turns a label into an exposure history
Conditions can begin, end, repeat or change during a study. A single label may hide the order and duration that explain the result. Preserve the start point, duration, sequence and relationship to measurement. State whether a setting applied during preparation, before measurement, during measurement or throughout the observation window.
Fictional teaching example: one group places an object in a setting for ten minutes; another uses the same setting for an hour. The condition name matches, but the histories do not. A broad search may retrieve both. A valid comparison still has to review duration and object state.
Recommended timeline records events separately from results. Avoid using a file's creation time as a substitute for exposure timing. If timing is unknown, state that. Students can practise by arranging event cards and asking which event preceded each observation. The activity reveals why “under condition A” is sometimes too compressed to support a scientific conclusion.
CHAPTER 7 OF 20 · Describe conditions precisely
7. Record material identity without inventing completeness
Some conditions involve a material, mixture or treatment. A name alone may not identify its composition, concentration, source or preparation. Preserve those details where they are known and relevant. Do not assume two materials are interchangeable because they share an everyday name or a broad category.
For a safe fictional exercise, compare paper strips stored in two named containers. If one container is open and the other closed, the container label is not the full condition. Describe the arrangement and the intended purpose. The example teaches contextual description without introducing unsafe chemical or biological experimentation.
Recommended metadata distinguishes material identity, amount, unit, batch and application procedure. Not every study needs every field, and unavailable detail should remain unavailable. This is a record-design principle, not an instruction to collect or handle materials. Specialist interventions require their own authorised protocols and qualified oversight; an ontology article only helps explain how meaning can be preserved in a legitimate record.
CHAPTER 8 OF 20 · Describe conditions precisely
8. Controls are defined by what they control
“Control group” is not a complete description. A comparator may account for handling, a carrier material, a baseline state or another part of the intervention. Explain which difference the control is intended to isolate. A shared XCO term does not establish that the control and test groups were otherwise comparable.
In a fictional paper study, moving only the test objects to another table could change handling and location as well as the intended setting. A comparator that receives similar handling may help separate those effects, but the record must explain the design. Do not call a group adequate simply because it was named control.
Recommended comparison tables show every planned difference and every relevant shared condition. If several things changed together, state that the outcome cannot distinguish their individual contributions without more evidence. For students, ask “What did the control actually remove from the explanation?” This encourages reasoning about the design rather than treating control as a word that automatically validates the experiment.
CHAPTER 9 OF 20 · Describe conditions precisely
9. Batches and sites are contextual groups
Objects measured on the same day, in the same room or by the same processing batch can share circumstances. Those groupings matter when interpreting variation. NIST notes that measurement variability across days and handling conditions can differ from short-term instrument precision. Context records help make those differences visible. Source: NIST variability.
Fictional case: every test object is measured on Monday and every control object on Tuesday. Any group difference is entangled with day. Recording the day reveals the problem; naming the condition correctly does not fix it. A revised design may need to distribute groups across days or otherwise address the comparison explicitly.
Recommended datasets preserve batch and site identifiers, including links between related observations. Do not treat many rows from one batch as if they represented many independent environments. This is a useful student insight: the unit that receives a condition and the unit recorded in a table may differ. Row count alone does not determine the strength of evidence.
CHAPTER 10 OF 20 · Describe conditions precisely
10. Randomisation, blocking and recording do different jobs
Recording conditions tells us what circumstances were present. Randomisation and blocking are design choices intended to address particular sources of variation. They cannot be reconstructed after the fact simply by attaching ontology terms. A context-rich dataset can still come from a weak design; a strong design still needs context-rich reporting.
NIST's design guidance includes comparative, screening and response-surface objectives and connects design choice to factors and resources. For this article, the practical recommendation is narrower: identify the intended comparison and consult suitable design guidance before collecting evidence. XCO should describe legitimate circumstances, not act as a substitute for planning. Source: NIST experimental-design selection.
For a classroom exercise, shuffle safe object labels before assigning positions and record the allocation rule. Then compare with a fictional case where all long objects were placed in one position. Discuss why a metadata field can reveal allocation but cannot undo it. Understanding that difference helps students distinguish documentation from design.
CHAPTER 11 OF 20 · Test comparisons and controls
11. Do not convert context into cause
A condition associated with an outcome is not automatically its cause. Several circumstances can change together, and a measured setting can also be a marker for something unrecorded. XCO makes circumstances describable; causal interpretation needs an appropriate design, assumptions and evidence. Keep the observed relationship separate from the proposed explanation.
Fictional example: objects at one table give larger readings than objects at another. The tables may differ in lighting, operator, object allocation or measurement time. A location field is useful, yet it cannot tell us which difference produced the result. The next scientific move is to identify alternatives and design a discriminating check.
Recommended reports use calibrated language: describe the comparison actually made, list unresolved differences and explain what further evidence would be needed. Students can practise replacing “X caused Y” with a more precise sentence when the design does not isolate X. That is not timid writing. It is writing that makes the evidence and its boundary easy to understand.
CHAPTER 12 OF 20 · Test comparisons and controls
12. A shared condition class is a search aid
Grouping conditions under shared concepts helps readers discover related studies. It does not establish matching values, durations or implementation. Two records can share a broad condition family while representing substantially different circumstances. Use the hierarchy for retrieval, then inspect the detailed record before deciding whether to combine results.
A recommended filtering sequence checks type first, followed by setting, unit, timing, material, method and observation context. Report why a candidate record was excluded from a particular comparison. Preserve it in the discovery set if it remains relevant to a broader question. This keeps the search route useful without turning relevance into equivalence.
For advanced learners, define a competency question and a few expected inclusions and exclusions. Run the query against a pinned vocabulary and dataset. An unexpected result can reveal either a mapping error or an assumption in the query. The exercise turns ontology use into testable reasoning rather than a decorative list of terms in a report.
CHAPTER 13 OF 20 · Test comparisons and controls
13. Worked case: one label, two durations
Fictional teaching case: Group A records a named storage condition for five minutes and Group B records the same condition for fifty minutes. Both then measure the same defined endpoint with the same method. A search by condition type returns both groups. A pooled analysis that ignores duration loses an important design distinction.
The responsible review retains duration and states the target question. If the aim is to compare durations, the difference is intentional and should be analysed as part of the design. If the aim is to compare another factor while keeping storage history fixed, the records do not meet that assumption. The same contextual difference has a different role under a different question.
Ask students to write both interpretations. They should see that no field is automatically nuisance or treatment in every study. Meaning depends on purpose. The recommended condition matrix makes that purpose explicit, while the XCO mapping organises the circumstance type. Neither replaces the source timeline that tells us what was actually done.
CHAPTER 14 OF 20 · Test comparisons and controls
14. Worked case: a misleading zero
Fictional dataset: one group has a monitored light level of zero in the chosen units; another group has no light record. Converting the missing entry to zero makes the table look complete, but changes unknown context into a specific measured state. The condition class cannot justify that conversion.
Keep planned setting, observed value and observation status separately. The missing record might reflect a failed monitor, an omitted field or an unreported source. Unless evidence distinguishes those explanations, use a broad missingness code. Do not choose the most convenient state for the graph.
Students can compare summaries with and without the invented zero, then explain why the changed calculation is not a new observation. The lesson is practical: completeness and truth are different properties of a table. A smaller, honestly described comparison can be more useful than a large one built on fabricated context. State what remains unknown and which question can still be answered despite it.
CHAPTER 15 OF 20 · Test comparisons and controls
15. Worked case: matching type, mismatched phase
Fictional case: both datasets report the same environmental setting, but one applies it during storage and the other during measurement. The setting values match. The histories do not. A comparison that treats both as exposure during measurement changes the meaning of one source record.
Review the timeline and mark the phase explicitly. Ask whether the property could have changed before the final reading and whether the measurement occurred under the stated setting. If the source does not answer those questions, flag the record rather than infer a continuous exposure. A matching label is a starting clue, not the final decision.
This case can be taught with event cards and no real intervention. Students arrange storage, transfer and measurement, then write a condition sentence for each timeline. The resulting sentences make the difference visible. Advanced readers can turn those sentences into separate event records linked to the measurement, preserving sequence instead of compressing everything into one timeless tag.
CHAPTER 16 OF 20 · Practise, learn and continue
16. Version context meaning as well as data
The XCO repository provides ontology files and a public process for requesting term changes. Preserve the vocabulary version used for annotation and the source context description. When definitions or hierarchy change, review whether old mappings still fit. A database update can change query results without changing the underlying experiment. Source: the RGD XCO repository.
Recommended migration tests check unresolved identifiers, obsolete terms, changed ancestors and unexpected changes in condition-group counts. Version the mapping rules separately from the ontology. A new source description and a new classification rule are distinct events and should remain distinguishable in the audit trail.
For students, keep the first condition description and the revised version after feedback. Explain what detail was added and whether the observation itself changed. This shows how scientific records improve: we can learn to describe evidence better without pretending that we performed a different experiment. The history protects both honesty and the ability to resume work later.
CHAPTER 17 OF 20 · Practise, learn and continue
17. Teach context through safe fair comparisons
Choose a paper-based or purely fictional task. Ask learners to define the outcome, describe the measuring method and list relevant circumstances. Give two groups different record cards, then ask whether their results can be compared. The learning objective is to recognise missing context, not to conduct an intervention involving people, animals or hazardous materials.
For introductory Science learning, focus on what changed and what stayed comparable. At Secondary level, add timing, grouping and possible alternative explanations. Advanced students can create a context dictionary and a reviewed ontology mapping. These are suggested learning routes, not a claim that XCO is a compulsory Singapore examination topic.
Use the Science Learning Hub to continue at the appropriate level. When a student struggles, identify the precise missing layer: value, time, role or source. A clear repair is more useful than repeating “control your variables” without explaining which circumstance must be documented and why it affects the question.
CHAPTER 18 OF 20 · Practise, learn and continue
18. Give parents a manageable conversation
A parent need not become an experimental-design specialist to help. Ask the student to explain one fictional result and name three circumstances that would matter for comparing it with another result. Then ask where those circumstances appear in the record. This connects spoken understanding to written evidence.
If the answer is incomplete, add one detail at a time. For example, supply duration and ask whether it changes the comparison. Avoid turning a single gap into a judgement about general Science ability. Context reasoning is learned through examples, revision and checking, not through pressure to produce a perfect answer immediately.
In a small-group activity, one learner states the intended condition, another reads the realised record and a third asks what could still explain the result. Rotate roles and keep disagreements tied to evidence. The happy outcome is a more intelligible investigation: everyone can point to what is known, what is missing and the next question that would make the comparison stronger.
CHAPTER 19 OF 20 · Practise, learn and continue
19. Connect context literacy to education and work
Context literacy draws together scientific observation, mathematical comparison and data organisation. It can help students understand why careful research needs more than a collection of numbers. The relevant capability is explaining how circumstances affect interpretation, not memorising a specialist acronym or claiming that a tidy dataset guarantees a correct conclusion.
For a portfolio, build a fictional condition matrix and show how three proposed comparisons pass or fail its stated criteria. Include timelines, missingness and a mapping rationale. Such a project can demonstrate thoughtful reasoning without collecting sensitive or risky data. Explain its scope and limitations as part of the finished work.
The Education Hub handles broader school and learning decisions; Careers by Subject Capability helps explore tasks associated with different interests. Real course, job and professional requirements must be checked separately against current official information. An XCO guide can open a useful conversation about research practice. It does not confer permission, qualification or guaranteed outcomes.
CHAPTER 20 OF 20 · Practise, learn and continue
20. Finish with an accountable context bundle
A recommended context bundle includes the study question, observation unit, condition types and releases, planned and realised settings, values and units, materials where relevant, timeline, controls, batch structure, deviations and source provenance. Link it to method and outcome records. Tailor the fields to the question and mark unknown information explicitly.
Write a readable summary: “These results were obtained under these circumstances; this comparison aligns these features and leaves these differences unresolved.” Check that the sentence matches the structured data. If it does not, repair the record before presenting a causal story or a pooled graph. Documentation should reduce confusion, not disguise it.
Why Science? Because context gives an observation a place in the world. XCO provides shared condition meaning; values, timelines and design judgement preserve the circumstances that make evidence useful. Start with one comparison whose context you can explain. Then extend the dataset carefully, keeping curiosity, responsibility and the route back to the source together.
Back to contents · Previous chapter · Continue to the Science Learning Hub
Sources and scope
Official and primary sources checked on 15 October 2026: OBO Foundry: Experimental Condition Ontology; RGD: official XCO repository; NIST: selecting an experimental design; NIST: measurement variability. Project descriptions above are attributed to their named sources. Worked datasets are fictional; record designs, review checklists and teaching activities are editorial recommendations, not official ontology requirements. No search-volume, clinical, admission or career-outcome claim is made.