A Capability Profile Without Labelling the Child | Evidence, Support Conditions and the Next Learning Move

A capability profile is a careful description of what a learner can currently do under stated conditions. It is not a personality description, a permanent ability label, a diagnosis, a prediction of the child’s future or a polished paragraph that turns one test result into an identity. A useful profile names the capability, the task, the support available, what was observed, what remains uncertain, the learner state, the next action and the evidence that would justify changing the route.

Parents often arrive with global language: “my child is weak in Mathematics”, “she is careless”, “he has no confidence”, “she cannot write”, “he is bad at exams”. Schools and tutors can unintentionally reinforce the same compression when a report, subscore or repeated comment is read as though it describes the whole learner. The eduKate Sengkang runtime does the opposite. It asks: weak at what, on which task, under what conditions, with how much support, based on what evidence, and how recently?

This page is the canonical learner-runtime owner for building a capability profile without labelling the child. It connects Assessment Evidence, Learning Runtime, Learning Practice & Review, How We Know Learning Has Really Held, the Tutor System and Learning Atlas V2.0. The profile exists to improve the next decision and should be revised or retired when new evidence changes that decision.

---

The one-sentence definition

Capability profile = a time-bounded, task-bounded evidence summary of what the learner can do, under which support and conditions, with stated uncertainty and a planned next test.

The phrase “time-bounded” matters because learners change. The phrase “task-bounded” matters because performance changes with the demand. The phrase “support and conditions” matters because an answer produced independently is different evidence from an answer produced after hints, modelling or collaborative work. The phrase “stated uncertainty” matters because educational evidence is rarely perfect. The phrase “planned next test” matters because a profile that cannot guide a next action becomes a label in disguise.

What a capability profile is not

Not a capability profile	Why it fails	Better question
“She is weak in Maths.”	The subject is too broad and no condition is stated.	Which mathematical capability fails first, on which tasks, and does it fail independently or only under pressure?
“He is careless.”	A behaviour word hides several possible causes.	Was the error copied, selected, calculated, represented, timed, checked or recovered incorrectly?
“She is a visual learner.”	A fixed learning-style label can distort task design and expectations.	Which representation helps this specific capability become understandable without changing the target?
“He is an A student.”	A score band is not a capability description.	Which capabilities are stable, which transfer, and which still depend on familiar conditions?
“She lacks confidence.”	Confidence may be a consequence rather than the cause.	What evidence does the learner have about success, failure, support and independent performance?
“He needs tuition.”	Support is prescribed before the problem is located.	What teachable bottleneck would tuition solve, and what evidence would show it can later be reduced?

Why labels are attractive—and why they are dangerous

Labels compress complexity. A parent can remember “careless” more easily than a six-step description of when a child loses accuracy. A tutor can say “weak foundation” more quickly than identifying the exact prerequisite. A school report can summarise a term in a few words. Compression is sometimes necessary, but it becomes harmful when the compressed phrase starts controlling future expectations.

A label can create a self-reinforcing route. If a learner is described as “weak”, adults may give easier work, more prompts and fewer opportunities to demonstrate independence. The learner then produces less independent evidence, which appears to confirm the original label. If the learner is called “strong”, adults may overlook fragile foundations or interpret requests for help as attitude rather than evidence. A capability profile interrupts that loop by forcing the description back onto observable conditions.

The safest educational language is provisional where the evidence is provisional. “On two unfamiliar ratio word problems this week, she needed a representation prompt before choosing a method” is less elegant than “weak in ratio”, but it tells us what happened, how much help was needed, what kind of task exposed the issue and what to retest.

The minimum fields of a useful capability profile

1. Capability. Name the smallest educational capability that still matters for the decision.
2. Task. State what the learner was asked to do.
3. Conditions. State timing, familiarity, format, distractions, collaboration and other relevant context.
4. Support. Record what prompts, hints, examples, tools or accommodations were available.
5. Observation. Record what the learner actually did, not what the adult thinks the learner “is”.
6. First weak link. Identify the earliest unreliable move supported by the evidence.
7. Learner state. Blocked, Fragile, Stable, Transfer-Ready or Examination-Ready for this capability.
8. Uncertainty. Name what the current evidence cannot yet establish.
9. Next action. Choose the smallest intervention justified by the evidence.
10. Retest. Specify fresh evidence that could confirm, refine or overturn the profile.

A compact profile template

Field	Example entry
Capability	Compare fractions with unlike denominators and justify the comparison.
Task	Four P5 fraction questions: two numerical, one bar model, one word problem.
Conditions	Untimed; mixed with decimals; one-week delay after teaching.
Support	No worked example. One generic prompt: “Show the quantities another way.”
Observation	Correct on the two numerical items; needed the prompt on the bar model; chose the wrong whole in the word problem.
First weak link	Representation of the whole changes when the context changes.
State	Stable on familiar numerical comparison; Fragile/Blocked at contextual representation.
Uncertainty	Only one contextual item sampled; cannot yet generalise to all fraction word problems.
Next action	Contrast two problems with different wholes; require learner explanation before calculation.
Retest	Fresh changed-context problem after three days, no prompt; then mixed set one week later.

The profile describes a capability, not the child

One learner can have several different states at the same time. A Secondary 2 student may be Transfer-Ready in routine algebra, Stable in graphs, Fragile in comprehension answer scope and Blocked when a Science question requires causal explanation. Calling the whole child “strong” or “weak” destroys information that the learning system needs.

This is also why the profile should avoid identity sentences such as “Alicia is Fragile”. The better sentence is “Alicia’s current retrieval of the algebraic factorisation pattern is Fragile after a one-week delay.” The state belongs to the capability under stated conditions. It can change. It may not apply to another subject. It should not be carried forward after the evidence has changed.

The five learner states inside a capability profile

Blocked

Blocked means the learner cannot reliably begin, interpret the demand or select a useful first move under the stated conditions. It does not mean the learner knows nothing. A learner can possess relevant facts yet still be blocked because the task representation, language or method-selection step fails.

A Blocked profile should identify what made a first move possible: a diagram, rephrased instruction, prerequisite reminder, worked example, vocabulary clarification or another support. The next action is not to celebrate completion after heavy help. It is to determine the smallest support that restores a usable route and then test whether that support can fade.

Fragile

Fragile means success is available but unreliable. The learner may perform immediately after teaching, with a familiar example, after a cue, or on the same day, then fail after delay or reduced support. A Fragile capability often looks stronger than it is because classroom practice contains many hidden reminders.

A Fragile profile therefore needs delayed evidence. If the learner can do the skill only while the lesson is warm, the next test should not be another near-identical question five minutes later. It should include a delay, less support or a mild change in context.

Stable

Stable means the learner can perform reliably in familiar conditions without excessive support. Stability is important but should not be mistaken for transfer. Topic-labelled worksheets, predictable question order and repeated examples can create strong familiar-condition performance while method selection remains untested.

A Stable profile should state what remains familiar. That sentence is the bridge to the next test. “Stable on direct ratio questions with a diagram provided” immediately suggests a next condition: remove the diagram, change the story, mix topics, or ask the learner to choose the representation.

Transfer-Ready

Transfer-Ready means the learner can recognise and adapt the capability when wording, representation, context or neighbouring demands change. This is not the same as succeeding once on one unusual question. Transfer evidence should sample more than one changed condition where possible.

The profile should record which changes have already been survived. “Transfer-Ready across numerical and word-problem forms, but not yet tested under time pressure” is more informative than “mastered”. It preserves the distinction between flexible use and examination performance.

Examination-Ready

Examination-Ready means the capability survives the additional interference of mixed papers: time, switching, fatigue, question order, checking, mark allocation and recovery after a difficult item. A student can be Transfer-Ready but not Examination-Ready if the skill disappears when the whole paper competes for attention.

The profile should therefore separate topic knowledge from paper behaviour. If knowledge is stable but the learner leaves reachable marks because of pacing, the next owner is Examination Craft rather than another chapter worksheet.

Support must be recorded because support changes the meaning of success

A correct answer is not self-interpreting evidence. If one learner solves the problem independently and another reaches the same answer after a targeted hint, the outputs match but the evidence does not. Both performances may be useful, but they support different conclusions.

Support should be recorded without shaming the learner. The question is not whether help was “allowed”. Legitimate support can be exactly what good teaching requires. The purpose of recording support is to know what the learner can currently carry and what should be tested next.

A practical support ladder

Support level	Example	What the result can show
0 · Independent	No prompt beyond normal task instructions.	Best evidence of what the learner can currently carry alone under these conditions.
1 · Generic orientation	“Read the question again.” “Show what you know.”	Whether a light attention cue restores the route without teaching the method.
2 · Strategic prompt	“Could a diagram help?” “What is the whole?”	Whether method selection becomes available after a targeted cue.
3 · Partial scaffold	A table, sentence frame, step sequence or partially completed representation.	Whether the learner can use the scaffold productively; not yet independent evidence.
4 · Model / worked example	Adult demonstrates a related problem or paragraph.	Evidence about imitation or guided understanding; weak evidence of independent use.
5 · Co-completion	Adult and learner jointly construct most of the response.	Useful teaching evidence but should not be reported as independent capability.
6 · Answer exposure	Correct answer or completed solution is shown.	Evidence can concern explanation after exposure, not original problem solving.

The ladder is not a moral hierarchy. Level 4 may be exactly the right teaching move for a Blocked learner. The error is to teach at Level 4 and then record the completed task as though it were Level 0 performance.

Conditions belong in the profile because performance is conditional

A task completed at home with unlimited time, notes open and a parent nearby is not the same condition as a closed-book classroom test. Neither condition is automatically “better”. They answer different questions. The capability profile should preserve those differences instead of flattening them into one mark.

Relevant conditions can include time limit, question familiarity, topic labels, order, availability of notes, calculator access, response mode, noise, fatigue, collaboration, device use, language load and whether the task is being observed. Record only the conditions that could reasonably change the interpretation.

Do not confuse access conditions with the learning target

Some learners require legitimate access supports: larger print, extra processing time, a quieter environment, assistive technology, a different response modality, clarified layout or another adjustment. A capability profile should not treat the presence of a legitimate accommodation as “cheating” or erase it from the record. The correct question is whether the learning target stayed the same and what the learner demonstrated under the access conditions that apply.

This distinction matters because silently removing an access support can make performance incomparable, while silently lowering the target can make success misleading. The profile must state both the target and the access conditions clearly enough that later evidence can be interpreted fairly.

Uncertainty is part of an honest profile

A profile should never pretend to know more than the evidence supports. If only one question was sampled, say so. If the learner was unusually tired, record that the observation may not represent typical performance. If a task mixed reading load and Mathematics, do not confidently attribute the error to Mathematics without further evidence.

Useful uncertainty statements include: “not yet tested after delay”, “only one unfamiliar item sampled”, “cannot separate language load from concept knowledge”, “performance may have been affected by heavy prompting”, “school accommodation conditions were not replicated”, or “the learner corrected the error after feedback but has not yet shown the correction independently”.

The first weak link keeps the profile actionable

A profile becomes useful when it identifies the earliest unreliable move that controls the rest of the task. In a multi-step Mathematics problem, the final incorrect answer may come from a representation error in the first line. In comprehension, the written sentence may be grammatically correct but answer the wrong scope. In Science, the keyword may be correct while the causal relationship is missing.

The first weak link is not always the first thing the learner did chronologically. It is the earliest point in the dependency chain where reliability breaks. Repairing that point should make later steps easier. If the later error remains after the earlier repair, the profile changes.

English worked example: comprehension answer scope

Suppose a Secondary 1 learner reads a passage accurately and can explain it orally, but loses marks on short-answer comprehension. A global label such as “weak comprehension” is too broad. A capability profile might show that the learner selects relevant evidence but writes beyond the question scope, adding an inference that the question did not ask for.

Task: four fresh questions requiring different answer scopes. Conditions: untimed, no model answers. Support: none initially. Observation: relevant evidence found in all four; scope exceeded in three. State: Stable in locating evidence, Fragile in controlling answer scope. Uncertainty: not yet tested under timed paper conditions. Next action: contrast narrow and broad question demands; learner marks the requested scope before answering. Retest: new passage after four days, then timed set.

That profile routes the learner differently from a student who cannot understand the passage at all. The same total mark could conceal completely different next actions.

Mathematics worked example: “careless” algebra

A Secondary 2 learner repeatedly loses signs while solving equations. The parent says “careless”. The profile begins by checking whether signs are lost during copying, simplification, transposition shorthand, distribution, or calculator entry. Across six fresh questions, the learner preserves signs when each operation is written explicitly but makes errors when using the phrase “move it to the other side”.

Capability: preserve equality while transforming an equation. Task: mixed linear equations. Conditions: untimed, no worked example. Support: prompt to write the same operation on both sides. Observation: accuracy rises immediately with the explicit equality representation. State: Fragile/Blocked in shorthand transformation; Stable when equality is represented. Next action: rebuild the transformation language; do not merely add more equations. Retest after delay, then mix with algebraic fractions.

Science worked example: keywords without mechanisms

A Primary 6 learner writes “evaporation” and “heat” in an open-ended answer but earns no mark. “Weak Science” would hide the useful distinction. The learner can define evaporation and identify the relevant process, but cannot connect the changed condition to the rate of evaporation and then to the observed result.

Capability: construct a cause → mechanism → observation explanation. Task: three unfamiliar setups. Conditions: untimed. Support: one prompt asking, “What changed first?” Observation: correct scientific term selected independently; causal chain incomplete without prompt. State: Stable in vocabulary, Fragile in mechanism construction. Next action: practise causal chains with varied contexts. Retest: changed setup after three days; later timed open-ended set.

Cross-subject worked example: the Secondary 1 transition

A child who was comfortable in Primary school begins missing homework and failing tests in Secondary 1. A subject-by-subject profile may initially show no single catastrophic knowledge gap. Instead, the learner is losing tasks, starting late, arriving at difficult work tired and relying on last-minute parental reminders.

The capability profile here concerns self-management: locate obligations, choose a first task, begin at a planned time, and close the loop by checking completion. The next owner is the Learning Runtime, not necessarily three new tuition classes. Subject profiles can then be added only where evidence shows genuine subject weak links.

The profile should be updated, not defended

A capability profile is a working model. Its value comes from being easy to change when better evidence arrives. If a learner succeeds independently on three delayed transfer tasks, the state should move. If a supposedly Stable capability fails on a simple fresh item, the profile should become more cautious. Adults should not defend the old profile because they wrote it.

The update rule is: new evidence can strengthen, narrow, split, merge or retire a profile. A broad “fraction problem solving” profile may later split into “representation of the whole” and “operation selection”. Two separate error profiles may merge when both are traced to the same language misunderstanding.

When to retire a profile

Retire a profile when the capability has become sufficiently reliable that special monitoring no longer changes the learning decision. Keep the history only where continuity genuinely matters. A learner should not carry an old fragility forever simply because it was once important.

Retirement is educationally important. It tells the learner that repair can end. It prevents adults from continuing to watch an old weakness so closely that every future mistake is interpreted through it.

The parent version should be shorter than the tutor version

Parents usually need the decision summary, not every observation. A useful parent-facing profile might contain five lines: what we are seeing; what the current evidence suggests; what remains uncertain; what we are changing next; and when we will review. That is enough to coordinate action without turning the child into a case file.

For example: “On unfamiliar fraction problems, Ciara can calculate once the representation is clear but sometimes chooses the wrong whole. We are treating this as a representation fragility rather than a general fraction weakness. For two weeks we will contrast changing-whole problems and then test a fresh problem without prompts. We will review after the fresh task, not from worksheet completion alone.”

The student version should preserve agency

A student-facing capability profile should help the learner recognise the next move. It should not read like an adult diagnosis. Replace “you are weak at inference” with “you usually find the right sentence, but when the question asks what the writer implies, you sometimes stop at the literal meaning. This week your job is to name the clue and explain what it suggests before writing the answer.”

The student should know what success will look like and what support is expected to fade. A profile becomes more powerful when the learner can challenge it with new evidence: “I think I can do this without the prompt now.”

The tutor version needs enough evidence for continuity

A tutor may need more detail because another lesson, another tutor or a later review must understand why the route was chosen. The tutor version should record sample tasks, support levels, error pattern, first weak link, current state, intervention and planned retest. It should still avoid unnecessary personal data and speculative labels.

The Tutor System uses the profile as a coordination object: parent context enters, learner evidence is gathered, tutor judgement chooses the next teaching move, and later independent performance decides whether the support should continue, change, reduce or stop.

Profile language that protects the learner

• Prefer: “On these tasks…” Avoid: “She always…”
• Prefer: “Current evidence suggests…” Avoid: “He is…”
• Prefer: “Needed a strategic prompt…” Avoid: “Couldn’t do it.”
• Prefer: “Not yet tested after delay…” Avoid: “Mastered.”
• Prefer: “The first weak link appears to be…” Avoid: “The problem is definitely…” when evidence is thin.
• Prefer: “Under the current access conditions…” Avoid: treating accommodations as noise.
• Prefer: “Retest on a fresh task…” Avoid: relying on the corrected original item.

How many samples are enough?

There is no universal number because tasks vary in complexity and consequence. One sample can justify a question; it rarely justifies a broad label. A practical educational profile often improves sharply once the tutor has several observations that vary at least one important condition: immediate versus delayed, familiar versus changed, supported versus independent, topic-labelled versus mixed, untimed versus timed.

The evidence threshold should rise with the consequence of the conclusion. A small decision such as “practise this representation tomorrow” can be made from modest evidence. A large conclusion such as “the child is not ready for this subject” deserves a much stronger and more varied evidence base.

Do not average away the pattern

Averages can hide state changes. A learner who scores 60%, 60%, 60% is different from a learner who scores 90%, 30%, 90%, 30% even when the mean is identical. The second pattern suggests instability or strong condition effects. The profile should preserve variance when variance changes the next action.

Subscores can also create false precision. If only two questions represent a category, a 50% subscore may be one item wrong, not a stable estimate of capability. Record the actual task evidence when the category is too small to support a broad conclusion.

Opportunity matters: absence of evidence is not always evidence of inability

A learner cannot demonstrate a capability that the task never required. If a worksheet supplies the diagram, it cannot tell us whether the learner can choose a diagram independently. If every practice set announces the topic, it cannot tell us whether the learner can select the method from mixed work. If the tutor always asks the leading question, it cannot tell us whether the learner would generate that question alone.

A capability profile should therefore include an opportunity check: did the learner actually have a fair chance to demonstrate the target capability under the stated conditions?

Group work needs individual evidence

In a small group, the best answer can hide individual uncertainty. A learner may agree with a peer’s correct method without being able to reproduce it. Group discussion is valuable, but the profile should distinguish shared performance from individual performance when the decision concerns one learner.

A quick independent exit item, a fresh explanation, or a later retrieval check can provide the individual evidence needed without dismantling collaborative learning.

AI and digital tools: assisted contribution is not later independent performance

AI, calculators, spelling tools, hints and digital platforms can be legitimate parts of learning. The profile must record what the tool contributed. A student who used an AI hint to identify the first move may have learned something important, but the successful assisted attempt is not yet evidence that the learner can generate the first move independently.

The clean sequence is attempt → support → completion → later fresh task with the intended support conditions. That preserves the value of assistance without confusing assistance with ownership.

Privacy: record the minimum educationally necessary information

A capability profile should not become a repository for unnecessary personal information. Record what affects the learning decision. Avoid speculative medical, psychological or family explanations when they are not needed and not supported. Sensitive information belongs only where there is a legitimate educational need and appropriate handling.

The profile should be shareable at the level necessary for continuity without exposing more than the educational job requires. “Needs a clear written task sequence during the current transition” may be enough; a long personal narrative may not be.

The complete profile cycle

1. Observe. Start from a real task or repeated symptom.
2. Bound. Name the capability and conditions narrowly enough to interpret.
3. Record support. Make hidden help visible.
4. Locate the first weak link. Separate cause from downstream error.
5. Classify provisionally. Use the five learner states for this capability only.
6. Choose the smallest intervention. Do not fix more than the evidence justifies.
7. Retest fresh. Prefer a new task over the corrected original.
8. Add delay or variation. Test whether the capability holds.
9. Update the profile. Strengthen, narrow, split, merge or retire it.
10. Return control. Fade support as the learner carries more of the route.

Where this owner sits in the eduKateSengkang system

Use Start Here when the family begins with an everyday symptom. Use How Assessment Evidence Works when the question is what a test or task can legitimately tell us. Use this page when the evidence needs to be assembled into a learner-specific capability model without turning that model into identity.

Then use Learning Practice & Review to choose the next practice and How We Know Learning Has Really Held to decide whether the repair survived delay, variation and transfer. Use the Tutor System when parent, student and tutor need to coordinate around the profile. Return to Learning Atlas V2.0 whenever the state or owner changes.

A final rule

The learner is always larger than the profile. The profile is a temporary instrument for choosing the next educational action. It should become more accurate with evidence, smaller as uncertainty resolves, and unnecessary when the learner no longer needs special routing for that capability.

Capability profiles across the school journey

A capability profile should become more sophisticated as the learner’s work becomes more sophisticated, but its logic does not change. In Primary 1, the profile may concern decoding a short sentence or making ten with counters. In Secondary 4, it may concern integrating several algebraic techniques under time pressure. In both cases the profile still asks what the learner did, under which conditions, with what support, what failed first, what remains uncertain and what evidence should come next.

The following school-stage examples are not model students or diagnostic categories. They are demonstrations of how the same evidence discipline can be used without turning a moment in learning into a fixed identity.

Primary 1: reading a sentence independently

A Primary 1 learner reads a familiar class reader fluently but hesitates on an unfamiliar decodable sentence. A weak profile says “good reader but lacks confidence”. A better profile identifies the actual conditions: familiar vocabulary, repeated text and teacher-supported practice versus a fresh sentence with one unfamiliar word.

Capability: decode and blend unfamiliar regular words inside a short sentence while preserving meaning. Task: six short sentences, four familiar and two fresh. Conditions: quiet one-to-one setting, no time pressure. Support: teacher waits five seconds before prompting. Observation: familiar sentences read fluently; fresh words decoded after first-sound prompt; meaning preserved once word was decoded. State: Stable on familiar text, Fragile on fresh decoding. Uncertainty: only two fresh items. Next move: short fresh-word retrieval over three days, then a new sentence with no prompt.

The profile avoids saying the child “is not a reader”. It identifies a narrow transition between familiar fluency and fresh decoding. That distinction protects both expectation and intervention.

Primary 2: arithmetic fluency versus number meaning

A Primary 2 student answers twenty addition facts quickly but becomes stuck when asked to explain why 8 + 7 can be thought of as 10 + 5. A speed-only interpretation might call the learner strong. A capability profile separates fact retrieval from decomposition and relational number sense.

Capability A: retrieve addition facts within 20. Capability B: decompose numbers to explain an efficient strategy. The first is Stable; the second is Fragile. The next move is not more speed practice. It is to make the decomposition visible with counters, ten frames and equations, then remove the representation and ask the learner to generate the strategy independently.

Primary 3: Science observation and inference

A Primary 3 student records “the plant is unhealthy” after looking at yellow leaves. The statement mixes observation and interpretation. The profile should not say “weak in Science process skills” after one item. It should record the specific boundary the learner crossed.

Task: observe three plant images and record only what can be seen. Support: none. Observation: two responses include interpretations rather than observable features. Follow-up prompt “Could a camera record that sentence?” helps the learner revise. State: Fragile in distinguishing observation from inference. Retest: new non-plant context after delay. The changed context matters because otherwise the learner may memorise “yellow leaves” rather than learn the observation rule.

Primary 4: comprehension inference

A Primary 4 learner finds literal details accurately but misses inference questions. The adult may say “inference is weak”. A useful profile is more precise: the learner notices the clue but does not connect it to prior sentence context, or the learner makes a plausible inference without citing sufficient textual evidence.

Those are different profiles. One requires relationship-building between clues; the other requires evidence discipline. A fresh passage should distinguish them before a large inference worksheet is assigned.

Primary 5: fractions, percentage and ratio

Primary 5 often creates broad parent concern because several multiplicative ideas expand at once. A child may appear weak across fractions, percentage and ratio even when one underlying issue controls all three: the learner does not reliably identify the reference whole.

A cross-topic capability profile can be justified only if the same weak link is observed in multiple task families. Sample one fraction problem, one percentage-of-quantity problem and one ratio comparison. If the learner selects the wrong whole in all three, the profile can cautiously state that reference-quantity identification is a shared dependency. If the errors differ, keep separate profiles.

Primary 6: topic capability versus PSLE execution

A Primary 6 student can solve percentage word problems in topic practice but loses those marks in a mixed PSLE paper. The profile should split topic competence from examination integration. Topic state may be Transfer-Ready; examination state may still be Fragile because the learner misclassifies the problem when no heading announces the topic.

The retest should therefore be a mixed set or paper section, not another percentage worksheet. This is a common point at which adults unintentionally test the wrong capability and then wonder why paper marks do not change.

Secondary 1: algebra language

A Secondary 1 student can substitute values into a formula but cannot form an expression from a verbal relationship. “Weak algebra” hides the distinction between symbol manipulation and algebraic translation. The profile should identify whether the learner understands variables, operation order, unknown versus changing quantity, and how language maps onto structure.

A fresh task can contrast “three more than x”, “three times x”, “three less than x” and “x less than three”. If the learner’s error tracks language direction rather than arithmetic, the intervention belongs at the language-to-symbol boundary.

Secondary 2: factorisation and structure

A Secondary 2 learner performs factorisation when the worksheet is titled “Factorisation” but fails to factor before simplifying an algebraic fraction. Topic practice suggests mastery; mixed application reveals that method selection is not yet stable. The capability profile should record both facts rather than average them into one score.

State: Stable for execution when method is signalled; Fragile for structural recognition when method must be selected. Next move: contrast cases where factorisation helps and where it does not; ask the learner to name the structure before calculating.

Secondary 3: argument quality in English

A Secondary 3 student writes long essays with many examples but receives modest marks. A global profile such as “weak writer” is especially damaging because it hides assets. The learner may generate relevant ideas and write fluent sentences while lacking hierarchy between claim, reason, evidence and explanation.

A capability profile can preserve strengths: idea generation Stable; paragraph evidence selection Stable; explanation of why evidence supports the claim Fragile; whole-essay prioritisation Fragile. That profile suggests a very different intervention from “write more essays”.

Secondary 3: Additional Mathematics readiness

A learner entering Additional Mathematics may perform well on current Secondary Mathematics but rely heavily on memorised procedures. A readiness profile should sample algebraic manipulation, equation solving, graph interpretation and ability to learn from a worked example—not merely use one Mathematics mark as a proxy for future A-Math performance.

The profile should also avoid becoming a gatekeeping identity. “Not currently ready on these dependencies” is a different educational statement from “not an A-Math student”. Readiness can change after targeted repair.

Secondary 4: examination recovery

A Secondary 4 learner sits a full paper, gets stuck on one early question and loses ten minutes. Later questions are rushed. The final score is lower, but the profile should not automatically downgrade all topic capabilities. The paper evidence shows an execution weakness: leave-and-return decision, state reset and time reallocation after difficulty.

The next test should preserve the relevant paper conditions. If the student succeeds on a short topic set, that does not answer whether recovery is improving. A timed mixed section with a deliberately difficult item may be the better retest.

One learner, several profiles: a complete worked case

Consider a hypothetical learner, Mira, in Secondary 2. Her parents report three symptoms: Mathematics marks are inconsistent, English homework takes too long, and she rarely asks for help. The temptation is to create one narrative—“Mira lacks confidence and organisation”—and make every observation fit it. A capability-profile approach resists that compression.

Profile 1: Mathematics algebraic transformation

Evidence across two class tests and one fresh tutor set shows that Mira solves equations accurately when she writes each operation explicitly. Errors increase when she uses compressed shorthand. Support: one prompt to preserve equality restores accuracy. Current state: Stable with explicit representation, Fragile with compressed notation. Next action: practise transformations with self-explanation, then fade written prompts. Retest after one week in mixed equations.

Profile 2: English homework duration

Observation shows that Mira spends long periods rereading the task and editing the opening sentences before the whole response is planned. Her language knowledge is not obviously weak. The profile concerns task initiation and planning: Blocked/Fragile at converting a broad writing prompt into a manageable plan. Next action: two-minute purpose/audience/content frame before drafting; adult does not supply ideas. Retest on a fresh prompt.

Profile 3: help-seeking

In small-group work, Mira waits silently when stuck but answers accurately after a peer asks the question she needed. The profile should not call her shy. Capability: identify and communicate the point of uncertainty. State: Fragile. Support: tutor models three help-seeking sentence stems, then asks Mira to choose one. Retest across subsequent lessons with prompts faded.

These three profiles may interact, but they should not be merged prematurely. A global confidence label would obscure the fact that Mathematics representation, writing-task planning and help-seeking have different evidence, owners and interventions.

Profile granularity: how small is small enough?

A profile should be narrow enough to change a decision but broad enough to matter. “Cannot answer Question 4(b)” is too narrow; “weak in English” is too broad. A useful middle level might be “selects relevant evidence but exceeds comprehension answer scope” or “recognises percentage structure but misidentifies the reference whole when the whole changes”.

Granularity can change as evidence improves. Begin broad when little is known, then split the profile if different failure mechanisms appear. Conversely, merge profiles if several symptoms are repeatedly traced to one shared dependency.

The difference between a profile and a checklist

A checklist records whether components are present. A capability profile interprets performance under conditions. A student may check every box on a chapter checklist because each skill was demonstrated once with support, yet remain Fragile after delay. The profile includes evidence quality, support and uncertainty, so it can distinguish “seen” from “held”.

Checklists are still useful when they are inputs rather than conclusions. A topic checklist can tell the tutor what to sample. It should not automatically decide state.

The difference between a profile and a report card

A report card summarises performance over a period, often across multiple tasks and teacher judgements. A capability profile is narrower and more operational. The report may tell us that Science is 68%; the profile may show stable factual recall, fragile mechanism explanations and strong data reading. The two forms of information can complement one another.

Parents should avoid converting a report-book comment directly into a permanent profile. Comments such as “needs to participate more” or “careless” are signals for investigation, not complete causal explanations.

The difference between a profile and a diagnostic test

A diagnostic test is a measurement event. The capability profile is the evolving model built from that event plus other evidence. One diagnostic test may be highly useful, but the profile should still state what the test sampled, which conditions applied and whether later evidence confirms the pattern.

This is particularly important for online diagnostic platforms that return detailed subskill labels. Fine-grained output can look more precise than the number of items actually supports. Read the item evidence before treating the subskill label as reality.

The difference between a profile and a medical or psychological diagnosis

Educational capability profiles do not diagnose medical, psychological or developmental conditions. They describe observed learning performance under stated educational conditions. If a learner has a formal diagnosis or accommodation, the profile may need to respect and record relevant access conditions, but it should not attempt to reproduce clinical judgement.

A tutor should also avoid reverse-diagnosing from behaviour. Slow reading, avoidance, low marks, restlessness or silence can have many educational and non-educational causes. The profile records what happened in learning and routes concerns appropriately when they exceed the tutor’s role.

Evidence strength: a simple hierarchy

Evidence pattern	What it can reasonably support	What it cannot yet support
One observed error	A question and a possible weak link.	A broad ability label or stable learner-state conclusion.
Repeated same error on similar tasks	A provisional pattern within familiar conditions.	Transfer or examination conclusions.
Fresh task after delay	Stronger evidence of retention.	Generalisation to very different contexts unless sampled.
Changed representation/context	Evidence toward transfer.	Paper-level reliability unless timed/mixed conditions are sampled.
Mixed timed paper evidence	Evidence about examination integration.	A claim that every topic weakness is caused by examination conditions.
Repeated independent evidence across conditions	A stronger capability profile and possible state change.	Permanent claims about the learner.

What counts as fresh evidence?

Fresh evidence is not simply a new page number. A question can be technically new while preserving every cue from the practice set. A useful fresh task changes enough surface detail that the learner must reconstruct the method, but not so much that a completely different capability is introduced.

Freshness can come from new numbers, new wording, changed representation, different context, mixed ordering or delay. The choice should match what the profile needs to test.

Retest design should follow the uncertainty statement

If the uncertainty is “not yet tested after delay”, add delay. If it is “cannot separate reading load from Mathematics”, reduce the reading load or present the same mathematical structure with clearer wording. If it is “success may depend on the diagram being supplied”, ask the learner to choose whether to draw one. The uncertainty statement should make the next test obvious.

The capability-profile review meeting

A short review between tutor and parent can follow a fixed sequence. First, state the current capability narrowly. Second, show two or three pieces of evidence. Third, name the support conditions. Fourth, state the current learner state and uncertainty. Fifth, describe the next action and retest. Sixth, state what evidence would lead to reducing support.

This structure prevents meetings from becoming long narratives about effort or personality. It also makes disagreement productive: if a parent sees different evidence at home, the question becomes what task and conditions produced that evidence, not whose impression is more authoritative.

When parent observations and tutor observations disagree

A learner may appear independent at home but dependent in tuition, or the reverse. The disagreement may be real and useful. Conditions differ: task difficulty, fatigue, parent prompting, peer presence, time of day and familiarity can all change performance.

Do not average the observations into a vague middle. Compare the conditions. Recreate a relevant task where possible. The difference itself may reveal the next weak link.

When school and tuition evidence disagree

School and tuition can sample different capabilities. A learner may score well on tuition topic work and poorly on school mixed assessments because the tuition set announces the method. Or a learner may perform modestly in tuition because the tutor deliberately uses unfamiliar transfer tasks while school homework is more familiar.

The capability profile should name the task differences before anyone concludes that one setting is “wrong”. The goal is comparability, not competition between institutions.

A profile should preserve strengths without creating a fixed ‘strength type’

Strengths matter because they can provide leverage. A learner with strong verbal explanation may use that strength to make mathematical reasoning explicit. A learner with strong visual representation may use diagrams to understand a Science system. But the profile should not convert a useful strength into destiny.

Write “currently explains causal relationships clearly in speech” rather than “auditory learner”. The first sentence is evidence-based and can change; the second invites a fixed teaching prescription.

Profiles for high-achieving learners

High marks do not remove the need for profiling. A high-achieving learner may have hidden fragility because familiar school tasks do not challenge method selection. The profile can ask whether success survives delay, changed conditions, ambiguity, open-ended tasks or explanation without rehearsed language.

The purpose is not to manufacture weaknesses. It is to ensure that enrichment is chosen from evidence rather than from a desire to make work harder for its own sake.

Profiles for learners who are currently struggling

When many capabilities appear weak, profiling prevents overload. Instead of generating a long deficit list, identify the first dependency that controls the largest number of current tasks. Stabilise it, retest, and then update the rest of the map. Some apparent weaknesses may disappear once the shared dependency is repaired.

This is especially important after a bad result. A learner should not leave the review carrying ten simultaneous diagnoses. The profile system should reduce uncertainty and prioritise action.

Profiles during school transition

Transitions—Primary 1, Primary 3, Primary 5, Primary 6, Secondary 1, Secondary 3 and Secondary 4—change task demand and environment. A capability that was Stable under the old conditions may look Fragile under new ones without having disappeared. The profile should record the changed condition before assuming regression.

For example, a Secondary 1 learner may still understand Mathematics but now lose performance because homework starts later after CCA and the student has not developed a new planning routine. The capability profile and runtime profile need to work together.

Profiles during examination season

Near major examinations, adults often compress every issue into the score. A capability profile can separate what is still teachable from what is now an execution issue. It can also prevent last-minute overreaction to one bad paper.

The profile should become increasingly paper-relevant as the examination approaches: mixed retrieval, timing, checking and recovery. But if fresh evidence shows a Blocked prerequisite, the system may need to step back temporarily rather than pretending paper practice will fix it.

A profile is not complete until it names the next owner

Every useful profile should route somewhere. English comprehension may route to the English system or SETC depth. Mathematics structure may route to the Mathematics system or BukitTimahTutor. Science concept depth may route to the Science Hub and canonical reference shelf. Timing and paper control route to Examination Craft. Cross-subject initiation, retrieval or transfer route to Learning Runtime. Recovery after a destabilising failure may route to eduKateYishun.

Naming the owner prevents the profile from becoming an archive entry. It keeps the diagnosis connected to action.

The capability-profile casebook

The following cases are deliberately varied. Their purpose is to show how an evidence profile changes the educational decision. Each case begins with the kind of sentence a parent, learner or tutor might naturally say, then replaces it with a more useful capability model.

Case 1: “She is careless in Mathematics”

Symptom: the learner loses marks in otherwise familiar Mathematics work. First assumption: carelessness. Evidence review: twelve errors across three papers. Three are copying errors, two are operation-selection errors, four are sign errors after compressed algebraic shorthand, two are unanswered final parts, and one is a calculator entry error. The word “careless” has merged at least five different mechanisms.

Profiles: copying accuracy is Fragile under time pressure; algebraic equality transformation is Fragile when notation is compressed; question-completion monitoring is Examination-Ready only inconsistently. The next route splits: algebra repair goes to Mathematics, final-part monitoring goes to Examination Craft, and the calculator error is monitored rather than overtreated after one occurrence.

This profile protects the learner from being told to “be more careful” when several concrete changes are available.

Case 2: “He understands in class but fails the test”

Symptom: teacher explanation makes sense, homework is completed, test score remains low. Evidence review compares three conditions: immediate guided practice, next-day independent retrieval, and mixed assessment. The learner performs strongly during guided practice, falls sharply after delay and then chooses the wrong method in mixed work.

Profile A: retrieval of the taught method is Fragile after delay. Profile B: method selection is Blocked/Fragile when the topic label disappears. The next route combines spaced retrieval with mixed practice. A new explanation of the same lesson is not the first intervention because understanding during explanation was not the missing condition.

Case 3: “She reads all the time but comprehension is weak”

Symptom: avid reader, modest comprehension marks. Evidence: literal questions are strong, inference is mixed, answer scope is repeatedly too broad, and vocabulary-in-context is strong. The learner’s reading volume is real but does not automatically produce exam-answer discipline.

Capability profile: passage understanding appears Stable; evidence selection Stable; answer-scope control Fragile; time use under paper conditions not yet tested. The next action is targeted answer-scope work rather than prescribing more leisure reading as the main remedy.

Case 4: “He knows the formula but cannot start”

Symptom: formula recall is strong, unfamiliar Mathematics questions produce blank starts. Evidence: when the tutor names the topic, the learner proceeds; when the problem is mixed, the learner cannot identify which quantities and relationships matter. A diagram prompt often unlocks the route.

Capability profile: formula retrieval Stable; representation selection Blocked/Fragile; transfer from wording to mathematical structure Fragile. The intervention is not another formula sheet. It is problem representation, contrast cases and method-selection practice.

Case 5: “She is losing motivation”

Symptom: Secondary 3 learner procrastinates and says schoolwork is pointless. Evidence: start times have become later, A-Math errors cluster around one unresolved algebra dependency, English remains stable, sleep has shortened, and the learner still engages strongly with CCA and project work.

The profile should not turn “motivation” into the whole explanation. Algebraic dependency is Blocked/Fragile; weekly workload control is Fragile; broad willingness to engage is not absent. The next route repairs the dependency, reduces unproductive repeated practice and rebuilds the weekly system. Motivation is monitored as an outcome rather than treated as a character defect.

Case 6: “He never asks for help”

Symptom: quiet learner waits silently when stuck. Evidence: the learner can identify the exact step of uncertainty when asked privately but does not initiate a question in the group. In written reflection, the learner can articulate the gap accurately.

Capability profile: metacognitive identification of uncertainty is Stable; public help-seeking initiation is Fragile. The intervention is not “speak more”. It is to practise low-cost help-seeking moves, agree on a signal, and fade the tutor’s invitation over time.

Case 7: “She is brilliant at home but the exam destroys her”

Symptom: high home-practice accuracy, volatile examination results. Evidence: home work is untimed and topic-labelled. In timed mixed papers, the learner overspends time on early difficult questions, performs little checking and makes late-paper errors on otherwise Stable topics.

Capability profile: subject knowledge Transfer-Ready in sampled topics; paper pacing and recovery Fragile; checking routine inconsistent. The next owner is Examination Craft. More topic teaching may increase workload without addressing the failing condition.

Case 8: “He is weak in Science”

Symptom: overall Science mark is low. Evidence decomposition shows strong multiple-choice factual recall, mixed data interpretation, weak causal explanations and a tendency to state keywords without relationships.

Profile: factual retrieval Stable; graph reading Fragile; mechanism explanation Blocked/Fragile. The next route focuses on mechanism chains and data-to-claim reasoning. “Weak Science” is retired because it is no longer useful.

Case 9: “She cannot write”

Symptom: slow writing and incomplete compositions. Evidence: oral storytelling is coherent, ideas are relevant, handwriting is effortful, transcription errors increase under speed, and sentence quality drops late in the task.

Profile: idea generation Stable; oral sequencing Stable; transcription fluency Fragile; extended written endurance Fragile. The intervention should not treat idea generation as weak. It may require shorter high-quality writing intervals, transcription practice, planning that reduces working-memory load, and a fresh extended task later.

Case 10: “He is not ready for A-Math”

Symptom: one school Mathematics result is lower than the parent hoped. Evidence review samples algebraic manipulation, equations, graphs, proportional reasoning, persistence on unfamiliar work and ability to learn from a worked example. Some dependencies are strong, one algebraic area is Fragile, and the learner improves quickly after targeted repair.

Capability profile: not yet ready on one specific dependency under current conditions; broader readiness remains open. The statement “not an A-Math student” is rejected because the evidence does not justify a permanent identity or pathway conclusion.

The three profile views: parent, student and professional

One evidence model can be expressed differently for different readers. The underlying facts should remain consistent, but the level of detail and language should match the job.

Parent view

Field	Parent-facing wording
Observed problem	“In mixed Mathematics papers, the same algebra sign error is recurring.”
Current interpretation	“The issue appears to be equality transformations when shorthand is used, not a broad algebra weakness.”
Support condition	“Accuracy improves when each operation is written explicitly.”
Next action	“For two weeks we will rebuild the transformation routine and then mix it back into full questions.”
Review evidence	“We will review from a fresh mixed set after a one-week delay.”

The parent view should help the family understand why the route was chosen and when it will be reviewed. It should not require the parent to interpret a technical case file.

Student view

Field	Student-facing wording
What is already working	“You usually know what equation step is needed.”
Where it breaks	“When you use the shortcut ‘move it across’, signs sometimes change without a valid operation.”
This week’s move	“Write the same operation on both sides until the equality stays visible.”
What independence will look like	“You can choose when to compress the working without losing the sign.”
Fresh check	“Next week you’ll get mixed equations without a heading or reminder.”

The student version should make the path inspectable and achievable. It should preserve evidence of what already works rather than presenting the learner as a collection of deficits.

Professional / tutor view

Field	Professional record
Capability	Preserve equality across algebraic transformations.
Samples	Two school scripts + six fresh mixed equations.
Conditions	Untimed tutorial; no worked example; topic label absent on fresh set.
Support	Strategic prompt to write identical operation on both sides.
Pattern	Errors cluster when compressed transposition shorthand is used.
State	Fragile under independent shorthand; Stable with explicit equality representation.
Uncertainty	Not yet tested under full-paper time pressure.
Intervention	Explicit equality transformations → faded written structure → mixed equations.
Retest	Fresh mixed set in 7 days, then paper section if stable.

The professional record exists for continuity and decision quality. It should be concise enough to update rather than so elaborate that nobody maintains it.

The profile update log

For capabilities that require more than one review, an update log can prevent adults from rewriting history around the latest result. The log should show what changed in the evidence and why the route changed.

Date / review	Evidence	Interpretation	Action
Review 1	Needs diagram prompt on unfamiliar fraction problems.	Fragile transfer; representation not independently selected.	Teach contrast between two possible wholes.
Review 2	Fresh problem solved after 3-day delay without prompt; second changed problem still wrong.	Improving but transfer not yet reliable.	Add varied-context pair; no increase in difficulty yet.
Review 3	Three changed-context items correct across one week.	Transfer evidence stronger.	Move to mixed timed set; reduce targeted practice.
Review 4	Mixed paper accurate but one long delay on first item.	Transfer-Ready; paper-selection speed still developing.	Route to Examination Craft, retire special fraction repair.

The log makes progress visible without pretending it is linear. It also shows when a subject repair should end and a different owner should take over.

How to write an uncertainty statement

Many profiles become labels because the writer omits uncertainty. A precise uncertainty statement narrows the claim and protects the next decision.

• “Only two fresh items were sampled; the pattern is provisional.”
• “The learner used a strategic prompt, so independent method selection remains untested.”
• “The task combined heavy reading and Mathematics; the source of the error is not yet separated.”
• “Performance improved immediately after feedback; delayed retention remains unknown.”
• “The learner used the school-approved access arrangement; performance without it is not the relevant comparison.”
• “Group discussion preceded the response; individual reconstruction will be checked later.”
• “The result came from one unusually long session; fatigue may have affected late-item accuracy.”
• “The current profile describes G3 Mathematics algebra and should not be generalised to other subjects.”

How to choose a next action from the profile

The next action should be the smallest move that can change the interpretation. A capability profile is not a licence to prescribe a large programme. If the uncertainty is narrow, the next test may be enough. If a weak link is well established, a short repair may be justified. If the capability is already Stable, variation rather than reteaching may be the right move.

Profile finding	Likely next move	Avoid
Blocked at first move	Clarify target, model or scaffold the missing first move, then fade.	Assigning a large independent set that repeats failure.
Fragile after delay	Spacing and retrieval with less support.	Interpreting same-day success as mastery.
Stable only on familiar format	Vary representation, wording or context.	More identical topic pages.
Transfer-Ready but weak in timed paper	Examination Craft: pacing, selection, checking and recovery.	Reteaching the entire topic.
Strong capability, low opportunity to demonstrate	Create a valid opportunity to show it.	Calling the capability weak because evidence is absent.
High support dependence	Fade prompts systematically and test fresh work.	Keeping successful scaffolds permanently because they produce correct answers.

How to distinguish intervention from measurement

Teaching and measuring often happen in the same lesson, but they should not be confused. Once the tutor gives a hint, models a method or corrects the misconception, the task has become instructional. The learner’s response after that point is useful evidence about learning from support, but it is not a clean measure of what the learner could do before support.

A later fresh item is therefore essential. Without it, a profile can overstate improvement because the only evidence is the corrected task.

Why corrections are weak evidence unless they travel

A beautifully corrected paper can show that the learner understood feedback in the moment. It does not automatically show that the learner will detect or avoid the same error next time. The capability profile should record the correction as an intervention event and reserve state change for fresh evidence.

This is one reason learners can maintain excellent correction files while marks remain unchanged. The file documents adult-supported understanding; the next paper tests independent transfer.

The fresh-item reserve

Where possible, keep some tasks unseen and untaught so that later checks still have diagnostic value. If every available item is used during teaching, the tutor loses the ability to distinguish learning from memory of the practice set.

The reserve does not need to be large. Two or three fresh items chosen deliberately can provide more useful evidence than another twenty familiar questions.

The delay ladder

Delay	Useful question
Minutes	Can the learner reproduce the move after immediate practice without the model visible?
Next day	Did the capability survive the end of the lesson and normal forgetting?
Several days	Is retrieval becoming durable rather than lesson-dependent?
One to two weeks	Does the skill remain available after competing school learning?
Later mixed revision	Can the learner still select and use it when the topic is no longer foregrounded?

Not every capability requires every delay. The profile should choose enough delay to answer the uncertainty that matters.

The variation ladder

Variation	What it tests
New numbers / details	Whether the learner memorised the exact example.
Changed wording	Whether language changes break recognition.
Changed representation	Whether the learner can move between diagram, equation, table, graph or prose.
Changed context	Whether the underlying structure is recognised in a new situation.
Mixed neighbouring topics	Whether method selection survives competition.
Open-ended task	Whether the learner can construct a route when the method is not pre-specified.

The support-fading ladder

Support should fade when evidence permits, not according to a fixed calendar. A possible sequence is model → partial model → strategic prompt → generic prompt → silent wait → independent attempt. Some learners will skip levels; others will need to return temporarily. The profile records which level currently preserves productive struggle without replacing thinking.

The goal is not zero support in every circumstance. The goal is the minimum support required for the learner to engage with the intended target, with growing ownership where independence is part of that target.

When not to fade support

Do not remove legitimate access accommodations simply to “see what happens”. Do not remove a safety instruction. Do not strip away the only representation that makes the target accessible if representation itself is not the skill being tested. Support fading should be tied to the educational construct, not to an ideology that all help is bad.

Capability profiles and accessibility

A fair profile distinguishes access from target. If the target is scientific reasoning, allowing typed rather than handwritten response may preserve the target for a learner whose handwriting imposes unrelated load. If the target is handwriting fluency, typing would change the construct. The same accommodation can therefore be appropriate in one profile and invalid in another.

Record the target first. Then record which access conditions were used. Only compare performances that are meaningfully comparable.

Capability profiles and language load

Many tasks are cross-loaded with language. A Mathematics word problem can fail because of vocabulary or syntax even when the mathematical structure is available. A Science explanation can fail because the learner understands the mechanism but lacks the language to express the causal chain. A capability profile should separate these where possible.

One method is to hold the conceptual structure constant while changing language load. Another is to ask the learner to represent the problem nonverbally. The goal is not to remove language permanently; it is to discover which component currently limits performance.

Capability profiles and working memory

Complex tasks can overload working memory even when individual components are known. The profile should not automatically conclude that the components are weak. Compare the learner on isolated elements and on the integrated task. If isolated performance is Stable but integration fails, the next intervention may involve chunking, external representation or sequencing rather than reteaching every component.

Capability profiles and fluency

Slow performance can mean several things: the method is not automated, the learner is checking carefully, retrieval is weak, the task is unfamiliar, or the learner is using an inefficient but valid strategy. Time alone is not the profile. Observe the process.

Fluency profiles should usually record both accuracy and method. Faster incorrect performance is not progress. Slower deliberate performance may be a necessary phase before a reliable method becomes efficient.

Capability profiles and confidence

Confidence should be treated as evidence about the learner’s expectation, not as a substitute for capability evidence. A learner can be highly confident and wrong, or capable and uncertain. Compare predicted performance with actual performance over time.

A useful profile might say: “Before fresh algebra items, the learner predicts failure despite 80–90% independent accuracy.” That suggests a calibration issue. The intervention can then build trust from visible evidence instead of giving generic encouragement.

Capability profiles and motivation

Motivation also needs decomposition. A learner may avoid work because it feels impossible, because the reward structure is poor, because the task is boring, because fatigue is high, because failure has become expected, or because the work seems unrelated to goals. A capability profile addresses the learning component while avoiding unsupported claims about the learner’s character.

Capability profiles and behaviour

Off-task behaviour can be a signal, but it does not identify its own cause. Compare task difficulty, clarity, support, group conditions, timing and recent success. If behaviour changes when the task becomes accessible or when the first weak link is repaired, the learning environment was part of the mechanism.

This does not mean all behaviour is caused by task design. It means the capability profile should test plausible educational explanations before turning behaviour into identity.

Capability profiles and high support from parents

At home, parents often provide invisible support: reminders, task selection, checking, rephrasing, reassurance, materials and time management. A student may appear independent because the support architecture is so familiar that nobody notices it.

The profile can make one support visible at a time. Do not remove the whole home structure suddenly. Choose the support that the learner is ready to carry next, then observe.

Capability profiles and tutor dependence

A learner can become efficient at being tutored. The student recognises the tutor’s facial cues, waits for confirmation, asks “is this right?” after every line or delays action until the tutor speaks. Correct work in that setting can overstate independent capability.

Profile the dependency itself: which prompts trigger movement, how long the learner waits, whether self-checking occurs before confirmation, and what happens on fresh work between lessons. Then fade support deliberately.

Capability profiles and peer dependence

Peers can support learning, but one learner may become the group’s method selector while others become implementers. A capability profile should sometimes ask each learner to reconstruct the group’s reasoning independently. The aim is not to eliminate collaboration; it is to ensure group success does not substitute for individual evidence where individual capability matters.

Capability profiles and repeated failure

After repeated failure, profiles should become more careful, not more global. “Failed three tests” tells us the consequence is serious, but it does not tell us whether the cause is one repeated dependency, several unrelated weaknesses, examination execution or unstable study conditions.

Use the history to prioritise stronger evidence gathering, not stronger labels. If failure has destabilised the learner enough that ordinary study no longer works, route recovery to eduKateYishun before adding more of the same intervention.

Capability profiles and rapid improvement

Rapid improvement can be real, but one breakthrough should not automatically retire the profile. Ask whether the improvement repeats, survives delay and transfers. At the same time, do not keep the learner trapped in the old state after strong evidence has changed. Profiles must be willing to move in both directions.

Profile review frequency

Review frequency should match the volatility and consequence of the capability. A newly repaired Blocked skill may be reviewed within days. A stable maintenance capability may need only occasional checks. High-stakes examination execution may be reviewed after each full paper during the final phase.

Avoid reviewing so often that every normal fluctuation triggers a new route. The profile exists to improve decisions, not to generate continuous surveillance.

The profile closure statement

When special routing is no longer needed, close the profile explicitly. A useful closure statement says what evidence justified retirement and where ordinary monitoring will continue. Example: “Fraction reference-whole profile retired after three independent changed-context sets across two weeks and stable mixed-paper performance. Future errors return to normal Mathematics review unless the pattern reappears.”

Closure tells the learner and adults that a past weakness is not a permanent identity.

Profile bank: English

English is especially vulnerable to over-broad profiling because “English” combines reading, vocabulary, grammar, comprehension, writing, oral communication, listening, answer scope, audience control and examination execution. A learner can be strong in several systems and weak in one. The capability profile should preserve that architecture.

Observed symptom	Possible capability profile	Useful next evidence
Reads fluently but comprehension marks stay low.	Literal retrieval may be Stable while inference, evidence selection or answer scope is Fragile.	Fresh passage with questions separated by demand type; compare oral understanding with written answer scope.
Composition has many ideas but modest score.	Idea generation may be Stable; selection, sequencing, development or editing may be Fragile.	Plan and write one shorter fresh piece; inspect where quality drops rather than scoring only the whole essay.
Grammar worksheet strong, editing weak.	Rule execution Stable when category is signalled; error detection/method selection Fragile in continuous text.	Fresh editing paragraph with mixed error types and no category labels.
Vocabulary lists remembered, writing still plain.	Definition recall Stable; active lexical selection/collocation Fragile.	Fresh paragraph requiring precise word choice without target words supplied.
Summary exceeds word limit.	Comprehension may be Stable; selection, compression or paraphrase may be Fragile.	Ask learner to identify points before writing; compare point selection with compression separately.
Oral response confident but vague.	Fluency Stable; evidence, development or response structure Fragile.	Fresh prompt with a requirement to justify and extend one answer.
Listening answer lost despite understanding audio.	Listening comprehension may be Stable; note selection or response conversion Fragile.	Replay-free fresh task; compare oral recall, note capture and written answer.

English profile example: inference versus answer scope

A learner answers an inference question with a plausible idea but includes two unsupported claims. Rather than “weak inference”, the tutor samples four new questions. The learner identifies the intended inference correctly in three but adds extra speculation in all four. Capability profile: inference generation Stable/Transfer-Ready; evidence-bounded answer scope Fragile. The intervention teaches constraint, not inference from scratch.

English profile example: composition planning

A learner writes vivid openings but endings collapse and plots become overlong. The profile samples a spoken story outline, a five-minute written plan and one 35-minute composition. Spoken sequencing is strong, but the written plan contains too many events and no decision about the central change. Capability: narrative event selection Fragile. Next move: one-sentence story spine and consequence map before drafting; retest with a fresh prompt.

English profile example: reading dislike

A learner says reading is boring and avoids books. A profile should first separate access from preference. Sample oral reading fluency, silent comprehension, vocabulary load, text length and interest. If short accessible texts are understood well but longer texts become effortful because decoding remains slow, the profile concerns fluency/endurance rather than a fixed dislike of reading.

Profile bank: Mathematics

Mathematics profiles should follow dependency and representation. Final answers hide where the route failed. The strongest evidence often comes from the first wrong line, the representation chosen, the method-selection decision and whether the learner can verify the result.

Observed symptom	Possible capability profile	Useful next evidence
Careless mistakes.	Could be copying, sign control, operation selection, calculator entry, incomplete checking or time pressure.	Classify actual errors across several scripts before choosing an intervention.
Can do sums, not word problems.	Arithmetic Stable; language-to-relationship translation or representation Fragile.	Same mathematical structure in numerical and contextual forms.
Knows formula, cannot start.	Formula retrieval Stable; representation/method selection Blocked or Fragile.	Mixed fresh questions without topic headings.
Algebra collapsed in Sec 2.	One or more prerequisites—signed numbers, equality, expansion, factorisation—may be Fragile.	Short dependency probe ordered from earliest prerequisite upward.
Topical tests strong, full papers weak.	Execution Stable with topic cue; selection, switching or paper control Fragile.	Fresh mixed set and timed paper section.
Fast but inaccurate.	Fluency may be high; monitoring/verification or conceptual control may be Fragile.	Compare accuracy under normal pace versus deliberately slower checked work.
Slow but accurate.	Method may be valid but inefficient, retrieval may be weak, or checking may be excessive.	Observe method path and time by step, not only total duration.

Mathematics profile example: model drawing

A Primary 5 learner solves routine part-whole problems when the model is provided but cannot decide how to draw one independently. Profile: interpretation of provided models Stable; model selection/construction Fragile. Next action: contrast three problem structures and require the learner to justify the model before calculating. Retest with a changed story and no diagram.

Mathematics profile example: trigonometry

A Secondary 3 learner memorises trigonometric ratios but selects the wrong ratio in diagrams. The first probe shows angle identification is Stable, right-triangle recognition Stable, side naming Fragile when the diagram is rotated. Profile: representation orientation dependence. Next action: varied-orientation diagrams before adding harder word problems.

Mathematics profile example: checking

A learner is told to “check” but simply rereads working. A profile should define checking capability: identify likely error points, choose a verification method and act on discrepancy. If the learner can substitute an answer back into an equation after a prompt but never chooses that verification independently, method availability is Stable while checking selection is Fragile.

Profile bank: Science

Science profiles need to separate factual knowledge, relationships, mechanisms, evidence, data interpretation, experimental reasoning and answer expression. Keywords are often visible, but the profile should identify whether the learner understands what connects those words.

Observed symptom	Possible capability profile	Useful next evidence
Memorises keywords but loses marks.	Vocabulary Stable; causal relationship or answer chain Fragile.	Fresh open-ended questions requiring the same mechanism in changed contexts.
MCQ strong, open-ended weak.	Recognition may be Stable; explanation construction or evidence use Fragile.	Ask learner to explain selected MCQ reasoning, then compare with open-ended response.
Knows experiment terms, plans unfair tests.	Vocabulary Stable; variable control and causal test design Fragile.	Give a new investigation question and require variable/measurement plan.
Reads graph but conclusion overclaims.	Data extraction Stable; evidence boundary Fragile.	New graph where correlation, range and uncertainty matter.
Can state fact, cannot apply it.	Concept recall Stable; transfer to unfamiliar system Fragile.	Changed apparatus or real-world context requiring same underlying mechanism.
Changes prediction after seeing result.	Prediction/observation boundary Fragile.	Pre-register prediction and reason before revealing evidence.

Science profile example: fair tests

A Primary 4 learner can define “fair test” and name independent/dependent variables after direct teaching, but in a fresh plant-growth investigation changes both water and light. Profile: terminology retrieval Stable; experimental control Fragile in application. Next action: compare pairs of investigation plans and ask which conclusion each plan can legitimately support.

Science profile example: graph interpretation

A Primary 6 learner accurately reads values but says “increasing temperature always causes faster evaporation” from a graph covering only three tested temperatures. Profile: data reading Stable; claim-boundary reasoning Fragile. Next action: distinguish observation within measured range from general causal claim; retest on different data.

Science profile example: mechanism

A learner states “the metal spoon gets hot because metal is a good conductor” but cannot explain energy transfer. Depending on syllabus demand, the keyword may be insufficient. Profile should specify the required explanation level: material property recognition Stable; thermal-energy transfer chain Fragile. Retest with a different conductor/insulator setup.

Profile bank: examination performance

Examination profiles sit above subject profiles. They ask whether available knowledge survives the constraints of the paper. The profile should avoid blaming timing for a knowledge gap or reteaching knowledge when the true problem is execution.

Observed symptom	Possible examination profile	Useful next evidence
Cannot finish.	Time allocation, question selection, slow retrieval, overchecking or one-question trapping may be the cause.	Mark where time is lost during a timed section, not just which questions are unanswered.
Rushes and never checks.	Pacing calibration or checking selection Fragile.	Short timed set with explicit planned checking reserve, then fade the reminder.
Full papers much worse than topical tests.	Switching, method selection, fatigue or recovery Fragile.	Mixed section followed by full-paper evidence.
One hard question derails the paper.	Leave-and-return and state reset Fragile.	Simulated section containing one intentionally difficult item.
Changes correct answers during checking.	Verification quality Fragile; checking may be driven by uncertainty rather than evidence.	Record why each answer is changed and whether the change has a concrete discrepancy trigger.
Prelims collapse after good WAs.	Whole-paper endurance, difficulty calibration or pressure condition may differ.	Compare item-level failure pattern rather than treating the total mark as one capability.

Examination profile example: unfinished paper

A student leaves the final eight marks blank. Timing log shows twelve minutes spent on one five-mark question that remained incomplete. The rest of the paper pace is reasonable. Profile: overall speed is not the main issue; leave-and-return threshold is Fragile. Next action: decision rule tied to progress and marks; retest in a timed mixed section.

Examination profile example: checking that harms performance

A learner uses the final ten minutes to recalculate every question, becomes anxious and changes two correct answers. Profile: checking routine lacks prioritisation and discrepancy-based action. Next move: target unanswered parts, copied values, units, sign-sensitive steps and answers with an identified reason for doubt. Do not teach “check more”.

Profile bank: learning-runtime capabilities

Some of the most important profiles are not owned by one subject. They concern how learning is started, retrieved, transferred, monitored, corrected and carried independently.

Runtime capability	What to observe	Common mislabel
Task initiation	How long before a useful first move appears; what prompts unlock it.	Lazy / unmotivated.
Retrieval	What returns without notes after delay.	Forgot everything.
Method selection	Whether learner chooses a route when topic labels disappear.	Doesn’t understand.
Error detection	Whether learner notices discrepancy before adult correction.	Careless.
Help-seeking	Whether learner can identify and communicate the exact stuck point.	Shy / dependent.
Planning	Whether learner converts broad work into ordered next actions.	Disorganised.
Transfer	Whether capability survives changed representation/context.	Can do only easy questions.
Recovery	Whether learner can re-enter after an error, bad question or bad result.	No resilience.

Runtime profile example: task initiation

A learner takes fifteen minutes to begin homework. Observation shows the delay disappears when the first task is already chosen, but remains when several assignments compete. Profile: work initiation itself may be Stable once a target is selected; priority selection is Fragile. The intervention is a two-minute prioritisation routine, not generic motivation advice.

Runtime profile example: retrieval

A learner says “I knew it yesterday”. Immediate practice is accurate; next-day retrieval fails; a hint restores the method. Profile: Fragile retention. Next route: spaced retrieval with smaller successful returns, not a harder worksheet.

Runtime profile example: self-correction

A learner can fix an error as soon as the tutor points to the line but rarely detects errors independently. Profile: correction capability Stable after localisation; error detection Fragile. The next move is to teach where and how to check, then give fresh work where the tutor does not point to the error.

Profile bank: tutor-use capabilities

Tutoring creates its own capability conditions. The profile should sometimes describe how the learner uses tutoring, because a student can understand the subject yet use support in a way that blocks independence.

Tutor-use symptom	Profile question	Next evidence
Waits for tutor confirmation after every line.	Can learner self-check before seeking confirmation?	Tutor delays response and asks for learner’s verification.
Brings only blank homework questions.	Was there a genuine first attempt and a located stuck point?	Require attempt/evidence before explanation.
Copies model then stalls on fresh item.	Is example use producing abstraction or imitation?	Fresh structurally related item with model removed.
Asks broad “I don’t get it.”	Can learner identify the first unclear step?	Help-seeking sentence stem, then fade.
Performs only in tuition.	Which prompts/environmental cues are carrying performance?	Between-lesson fresh task with recorded support conditions.
Parent reports progress but learner cannot reproduce it.	What evidence is parent seeing and under what help?	Shared fresh independent task or school evidence review.

Tutor-use profile example: confirmation dependence

A learner completes algebra correctly but looks at the tutor after each line. When the tutor stays neutral, the learner slows and changes correct work. Subject capability may be Stable; confidence calibration and self-verification are Fragile. The tutor should not respond by offering more reassurance after every line. Teach a self-check and delay confirmation.

Profile bank: parent-support capabilities

Parent support can be described with the same evidence discipline. The goal is not to evaluate parenting. It is to make the learning conditions visible enough that support can be adjusted deliberately.

• Does the parent select the task or does the learner choose from known priorities?
• Does the parent rephrase every difficult instruction?
• Does the parent check completion or check correctness?
• Does the learner ask for help after an attempt or before beginning?
• Does the parent supply examples that become the learner’s answer?
• Can reminders be reduced without the system collapsing?
• Does the child perform differently when the parent leaves the room?

These observations can help identify support dependence without moralising. A parent who has been carrying the system often did so because it was necessary earlier. The profile helps decide what can now transfer back to the learner.

A reusable one-page capability profile

Capability: ____________________
Task sampled: ____________________
Date / recency: ____________________
Conditions: ____________________
Support available: ____________________
What the learner did: ____________________
First weak link: ____________________
Current learner state: Blocked / Fragile / Stable / Transfer-Ready / Examination-Ready
What remains uncertain: ____________________
Next action: ____________________
Fresh retest: ____________________
Owner / route: ____________________
Review / retirement condition: ____________________

The template is intentionally compact. The quality comes from the evidence, not from filling every line with jargon. If one field cannot be answered honestly, that gap may tell you what evidence to collect next.

A five-minute parent profile

1. Choose one repeated problem, not the whole subject.
2. Find one recent piece of work where it happened.
3. Ask what the learner could do before help arrived.
4. Write down the exact help that changed the attempt.
5. Decide whether the issue is start, knowledge, retrieval, selection, transfer, timing or checking.
6. Choose one small next action.
7. Choose when and how you will look for fresh evidence.

A parent does not need to diagnose every mechanism correctly. The purpose of the five-minute profile is to replace global worry with a testable question that can be taken to the learner, school or tutor.

A ten-minute tutor profile after a lesson

1. Save one representative error or successful fresh item.
2. Record support level used.
3. Name the earliest unreliable move.
4. State the current learner state provisionally.
5. Write one uncertainty statement.
6. Choose the next lesson action.
7. Reserve a fresh item or delayed check.
8. Decide what support should be faded if the retest succeeds.

A student self-profile

Students can profile their own learning without turning it into self-judgement. A simple self-profile asks: What could I start alone? Where did I get stuck? What hint changed the route? Could I explain why it worked? What will I try before asking next time? What fresh question would prove I can now do it?

This turns reflection into action. “I am bad at algebra” becomes “I can expand brackets, but I do not always recognise when factorisation is the first move. On the next mixed set I will name the structure before I calculate.”

Profile handover between tutors

When a learner changes tutor, a capability profile can prevent a total reset. The handover should not be a thick history. It should identify current high-leverage profiles, evidence, support conditions, what has already been tried, what changed, and the next planned retest.

The receiving tutor should still verify the profile. Similar symptoms can have different causes, and the new setting may change performance. A handover is a starting model, not an instruction to reproduce the previous tutor’s conclusions forever.

Profile handover from Primary to Secondary

The Primary-to-Secondary transition is a common place for old labels to become harmful. “Good at Maths”, “weak writer”, “careless” or “needs lots of help” may have been reasonable shorthand in one environment but become inaccurate as task demands change.

A better handover states portable evidence: “Stable with multi-step arithmetic; Fragile when selecting representations in unfamiliar word problems; usually begins homework independently but needs support prioritising several deadlines.” The Secondary environment can then test those capabilities again rather than inheriting identities.

Profile handover into examination season

As examinations approach, profiles should be compressed around the capabilities that can still change the result: unresolved knowledge dependencies, retrieval, transfer, pacing, checking and recovery. Low-priority historical weaknesses should not occupy the final weeks simply because they appear in old notes.

The exam-season profile should also state what no longer needs intervention. Protecting strong capabilities is part of a good plan.

Profile handover after a bad examination result

After a bad result, do not rewrite the profile from the total score alone. Compare the pre-exam model with the script. Which predicted fragilities actually appeared? Which supposedly stable capabilities failed? Which strong capabilities survived? Which new failure pattern emerged?

If the result destabilises the learner beyond ordinary repair, the profile can hand off to the eduKateYishun recovery system. When the learner is ready to re-enter ordinary learning, the Sengkang profile should be rebuilt from fresh evidence rather than simply restoring the old labels.

Common profile failure mode 1: adjective inflation

Profiles often drift from evidence into adjectives: weak, careless, lazy, bright, anxious, mature, distracted, gifted, shy. Some descriptors may be reasonable in ordinary conversation, but they should not substitute for capability evidence. Ask what observable learning behaviour the adjective is standing in for.

Common profile failure mode 2: profile accumulation

If every mistake creates a new profile, the system becomes unmanageable. Keep only profiles that change a current decision. Low-frequency or low-cost errors can remain in normal review. Merge related profiles when one shared dependency explains them.

Common profile failure mode 3: old evidence dominating

A learner may still be treated as weak in a capability repaired months ago because the original failure was memorable. Add review dates and retirement rules. Recent fresh independent evidence should outweigh stale impressions unless the old risk has a strong reason to remain on watch.

Common profile failure mode 4: score substitution

A score can trigger investigation but should not become the profile. “72%” does not tell the next move. Item patterns, conditions and support do. Even a detailed subscore requires task evidence before it becomes a capability conclusion.

Common profile failure mode 5: support blindness

A learner completes work correctly after extensive cues; the adult records “understood”. The next independent task fails; everyone is surprised. Record the support. Correct assisted performance is useful teaching evidence, but it answers a different question from independent performance.

Common profile failure mode 6: intervention loyalty

Adults can become attached to an intervention that worked once. The profile should be loyal to evidence, not the method. If the learner no longer needs the scaffold, fade it. If the scaffold is not producing fresh independent gains, change it.

Common profile failure mode 7: making the profile public identity

Avoid casually repeating the profile as the learner’s defining story—especially in front of the learner. “She is our fragile reader” turns a temporary state into social identity. Keep the language task-specific and update it openly when evidence changes.

Common profile failure mode 8: confusing effort with capability

Effort matters, but high effort can coexist with a poor method and low visible effort can coexist with strong automaticity. Profile the learning operation separately from the effort judgement. Then decide whether effort itself needs a route.

Common profile failure mode 9: profiling the wrong construct

A task may appear to test Mathematics while heavy reading controls success, or appear to test Science while memory of specialised vocabulary dominates. Clarify what the task is supposed to measure and which secondary demands are contaminating the observation.

Common profile failure mode 10: no exit condition

A profile without a retirement condition can live forever. State what evidence would allow the route to return to normal monitoring. This turns the profile from a label into a temporary control instrument.

Construct validity: make sure the profile is describing the skill you think it is

A capability profile can be carefully written and still be wrong if the task is not actually measuring the intended capability. This is a construct-validity problem. A Mathematics task may be dominated by reading load. A Science task may look like causal reasoning but really depend on remembering one rare term. A writing task may appear to measure ideas while handwriting speed controls how much can be produced.

Before interpreting failure, ask what the learner had to do besides the target skill. Secondary demands are not automatically unfair; real school tasks often integrate several abilities. The profiling job is to know whether the secondary demand is part of the target or a source of contamination.

Target capability	Possible contaminating demand	Useful comparison
Ratio reasoning	Dense English wording	Present the same ratio structure with simpler wording, then compare.
Science mechanism	Specialised vocabulary recall	Allow the key term but require the causal chain; then retest vocabulary separately.
Essay argument	Handwriting speed	Compare a shorter handwritten argument with an oral or typed planning explanation where appropriate.
Mental arithmetic	Task-switching and timing	Sample arithmetic separately before attributing full-paper errors to number facts.
Reading inference	Background knowledge unfamiliarity	Use passages with different knowledge demands while preserving inference structure.
Graph interpretation	Axis-reading conventions	Check basic graph-reading separately before judging higher-order inference from data.

The comparison does not “make the task easier” by default. It helps identify which component is controlling the result. Once the weak component is located, the full integrated task must return later because school performance ultimately requires integration.

Comparability: when two performances can and cannot be compared

A profile often relies on change over time, but change is meaningful only when the performances are comparable enough. A learner who scores higher on an easier worksheet has not necessarily improved. A learner who scores lower on a harder transfer task has not necessarily regressed. The profile should record the task and support differences before interpreting the score movement.

Comparability does not require identical tasks. In fact, identical tasks can overstate learning through memory. The goal is comparable construct demand with enough freshness to test the capability. When difficulty changes deliberately, say so and interpret accordingly.

Four kinds of comparison

• Same-demand fresh comparison: different surface details, similar underlying demand. Useful for checking retention.
• Support-faded comparison: task demand remains similar while prompts reduce. Useful for independence.
• Transfer comparison: representation or context changes while underlying structure remains. Useful for flexible use.
• Examination comparison: capability enters mixed, timed conditions. Useful for paper reliability.

Baseline: what could the learner do before the intervention?

Without a baseline, improvement is easy to overclaim. The baseline need not be a formal pre-test. It can be one or two representative fresh tasks, a marked school script, a timed sample or an observed explanation. The key is to know what support was present and what the learner could do before the new intervention started.

If a baseline is impossible because teaching has already occurred, say so. A profile can still be useful, but claims should be modest: “current performance after two guided lessons” is different from “improved because of the lessons”.

Intervention fidelity: did we actually run the repair we think we ran?

When progress is slow, adults often conclude that the intervention failed. Before changing routes, verify that the intervention was actually implemented as intended. Did the learner complete the planned delayed retrieval? Were prompts really faded? Did the tutor use fresh tasks or repeatedly revisit the same examples? Did home practice happen under the recorded conditions?

A route that was only partially implemented should not be judged as though it received a fair test. At the same time, if the route is too cumbersome to run consistently, that is itself evidence that the intervention may not fit the learner’s real week.

Intervention cost belongs in the profile decision

Every intervention consumes something: time, attention, travel, family coordination, learner effort, tutor time or opportunity to do other work. A technically effective intervention can still be a poor choice if its cost is disproportionate to the gain.

Capability profiling therefore asks not only “does this help?” but “does this help enough, under realistic conditions, to justify continuing?” A ten-minute retrieval routine that stabilises a recurring weak link may be more valuable than a two-hour weekly programme that duplicates school content without increasing independence.

Profiles in a three-student tutorial

Small-group tuition creates a useful middle ground between individual tutoring and a full classroom. Three learners can share explanation and discussion while the tutor still gathers individual evidence. But the group can also hide differences if the fastest student answers first or one learner becomes the default explainer.

A capability-profile routine in a three-student group can use four moments: shared model, individual first attempt, short comparison/discussion, and individual fresh exit item. The shared phase supports learning; the individual phases preserve evidence.

Three learners, one topic, three different profiles

Suppose three Secondary 2 students are learning algebraic factorisation. Student A factors accurately but cannot recognise when factorisation is useful in algebraic fractions. Student B chooses factorisation correctly but makes sign errors. Student C is blocked at common-factor extraction. One worksheet score could place all three near the same mark, but their next actions differ completely.

A three-student class should therefore share instruction where the job is common and split practice where the evidence differs. Precision does not require three separate lessons; it requires the tutor to know when the common route ends.

Profiles and lesson planning

A capability profile should change lesson planning. It should tell the tutor what to model, what not to reteach, what support to fade, what fresh task to reserve and what evidence to collect. If the profile does not alter the lesson, it may be too vague to be useful.

Profile state	Lesson planning implication
Blocked	Make the missing first move visible; reduce irrelevant load; create an achievable first independent step.
Fragile	Prioritise retrieval, spacing and reduced support before increasing difficulty.
Stable	Introduce controlled variation; require method selection rather than more identical execution.
Transfer-Ready	Use mixed and changed contexts; begin integrating paper-like conditions where relevant.
Examination-Ready	Maintain through selected full-paper evidence, checking and recovery; avoid unnecessary reteaching.

Profiles and homework selection

Homework should follow the profile rather than defaulting to “more of the chapter”. A Blocked learner may need a small scaffolded set. A Fragile learner needs delayed retrieval. A Stable learner needs variation. A Transfer-Ready learner may need mixed work. An Examination-Ready learner may need selective maintenance rather than volume.

The profile also tells the tutor what not to assign. Twenty identical questions can strengthen execution but hide method-selection weakness. A full paper can overwhelm a learner whose prerequisite is still Blocked. A hard enrichment set can destabilise a recently repaired capability before it has become reliable.

Profiles and feedback

Feedback should target the profile’s next move. If the profile concerns answer scope, feedback should help the learner notice the boundary of the question. If it concerns method selection, feedback should not simply demonstrate the method; it should expose the cue that should have triggered the choice. If it concerns checking, feedback should require the learner to select a verification route.

The strongest feedback produces a better fresh attempt. A profile can therefore record not only whether feedback was given, but whether the learner used it independently later.

Profiles and praise

Praise can also become a label. “You are brilliant at Maths” sounds positive but can make later struggle feel like identity failure. A capability-profile approach praises observable learning behaviour: “You noticed the representation was wrong and rebuilt it before calculating,” or “You checked the claim against the data instead of guessing.”

This keeps praise connected to processes the learner can repeat while still acknowledging genuine achievement.

Profiles and difficulty calibration

A learner can appear Blocked because the task is far beyond current readiness, or appear Stable because the task is too easy. Profile interpretation therefore requires some calibration of task difficulty. The question is not whether the learner succeeded, but what the success or failure reveals at this level of demand.

When difficulty is uncertain, sample a small ladder rather than one dramatic question. Find the point where independence changes. That boundary is often more informative than the hardest item completed.

Profiles and curriculum progression

School curricula are cumulative but not perfectly linear. A current chapter may depend on knowledge from years earlier. Capability profiles should therefore cross year boundaries when the evidence requires it. A Secondary 3 graph error can originate in Primary ratio or Secondary 1 algebra; a Primary 6 writing weakness can originate in sentence control from much earlier.

The profile follows the dependency, not the textbook page. This is one reason the Atlas routes by first weak link rather than by current chapter alone.

Profiles and subject transitions

New subjects or subject forms create temporary uncertainty. Primary 3 Science, Secondary 1 algebra, Secondary 3 Additional Mathematics and upper-Secondary argument writing all introduce new ways of representing knowledge. Early errors should be profiled provisionally until the learner has had enough opportunity to learn the new form.

The profile should distinguish “not yet taught”, “taught but Blocked”, “recently taught and Fragile”, and “previously Stable but now failing under a new condition”. These are different educational states even if the immediate answer is wrong in all four cases.

Profiles and revision planning

A revision plan based on profiles is different from a topic checklist. It allocates time according to state and leverage. Blocked prerequisites get repair. Fragile skills get repeated returns. Stable capabilities get mixed maintenance. Transfer-Ready capabilities get paper integration. Examination-Ready capabilities get lighter maintenance unless evidence changes.

This prevents weak topics from consuming all revision time while strong topics decay through neglect. It also prevents students from repeatedly practising their favourite stable work because it feels productive.

Profiles and examination countdowns

As a major examination approaches, the profile review window should shorten and the intervention set should narrow. The question becomes what can still change performance without destabilising what already works. Large new systems introduced late can carry more cost than benefit.

Near the examination, profiles should prioritise high-leverage knowledge gaps, reliable retrieval, method selection, timing, checking and recovery. Low-value historical weaknesses can wait.

Profiles and post-exam review

After the paper, do not immediately update the profile from memory of how the exam felt. Wait for reliable evidence where possible. When the script or result is available, compare the predicted profile with actual failure points. This calibration improves future judgement.

A profile that predicted fragile time management but the script shows early knowledge failures needs revision. A profile that predicted strong content but weak checking and the script confirms avoidable late-paper losses gains credibility.

Profiles and uncertainty over long periods

Some capabilities remain uncertain because they are rarely sampled. That is acceptable. Do not manufacture confidence by forcing a state label. Use “not enough evidence” when necessary and design a small opportunity to observe the capability if the decision matters.

The five learner states are tools, not compulsory boxes. An honest unclassified profile is better than a precise-looking but unsupported label.

Profiles and conflicting indicators

A learner may score well but require heavy support; perform slowly but accurately; speak confidently but write weakly; fail one difficult test after months of stable work. Conflicting indicators should trigger decomposition, not averaging. Ask which indicators refer to different capabilities or conditions.

The profile can hold more than one truth: “Stable accuracy with low speed”, “strong independent oral reasoning with Fragile written organisation”, or “Transfer-Ready content with Fragile examination pacing”.

Profiles and teacher comments

Teacher comments are valuable context, especially when they describe repeated classroom behaviour, but they should be translated into testable capability questions. “Needs to be more attentive” might lead to checking instruction-following, task initiation, note capture or sustained independent work. “Needs to elaborate” might lead to explanation depth or evidence use.

Do not discard the teacher’s observation; operationalise it.

Profiles and report-book trends

A sequence of marks can show direction but not mechanism. A falling trend justifies closer investigation. A stable average can hide changing strengths and weaknesses. A rising trend can still conceal support dependence. Use the report book as a signal source, then return to task evidence.

Profiles and parent anxiety

A capability profile can reduce anxiety when it narrows a broad fear into one actionable question. It can also increase anxiety if every normal fluctuation becomes a new entry. The system should therefore remain bounded: profile what changes a decision, ignore noise, and retire profiles when special action is no longer required.

Profiles and the language of ‘potential’

Avoid claiming that a profile measures a learner’s potential. It measures current performance under stated conditions. Potential is broader, future-facing and shaped by opportunities, teaching, development and many factors the profile cannot observe. A low current state should never be presented as a ceiling.

Profiles and capability growth

The most useful long-term picture is not a permanent profile but a sequence of state changes: Blocked → Fragile → Stable → Transfer-Ready → Examination-Ready, sometimes with reversals when conditions change. The history shows which supports helped and which conditions still matter.

Growth is therefore visible as increased range, reduced support and stronger survival under delay and change—not simply a higher score on the same task.

Profiles and autonomy

A mature capability profile gradually becomes unnecessary because the learner can perform the profiling moves internally: notice where they are stuck, distinguish knowledge from execution, choose a practice type, ask for targeted help, check fresh evidence and update the plan.

The deepest purpose of the profile system is to transfer diagnosis itself toward the learner.

Profile governance: who can change the learner model?

A capability profile should not become a document that only one adult is allowed to interpret. The tutor may maintain the working record, but the learner, parent, school evidence and fresh task performance can all challenge the model. Good governance means the profile changes when stronger evidence arrives, regardless of who produced the original description.

At the same time, not every opinion has equal evidential weight. A parent may have important home observations, but one general impression should not outweigh repeated fresh task evidence. A tutor may have detailed lesson notes, but those notes should not overrule a school result without examining why the conditions differed. The profile should privilege relevant evidence while still respecting context.

The learner’s right to challenge the profile

Where developmentally appropriate, the learner should be able to say, “I can do this now,” “that prompt is no longer necessary,” or “the problem is different from what you think.” The response should be neither automatic acceptance nor dismissal. Turn the challenge into a fresh test.

This is one of the strongest protections against labels. The learner is not trapped by an adult description; new evidence can revise the model.

The parent’s role

Parents contribute longitudinal context: when the problem began, whether it appears at home, how much prompting occurs, what the weekly load looks like, and whether the learner behaves differently outside tuition. The parent should not be asked to diagnose technical causes alone. Their strongest contribution is often observable context and change over time.

The tutor’s role

The tutor converts educational evidence into a testable teaching route. That includes selecting appropriate tasks, controlling support, identifying the first weak link, designing the retest and fading intervention. The tutor should keep causal claims provisional where evidence is mixed.

The school evidence role

School assessments, teacher comments, classwork and subject-level information can extend the evidence base. The capability profile should not copy school labels uncritically, but it should also not ignore school evidence because tuition performance looks better. Differences between settings may be diagnostically useful.

When different adults disagree about the learner

Disagreement is often a condition problem in disguise. A parent says the child “can do it at home”; the teacher says the child “cannot do it in class”; the tutor sees mixed evidence. Instead of choosing a side, describe each task, support condition and context.

Source	Observed success	Conditions to check
Home	Homework completed accurately.	Was the task selected by the parent? Were notes open? Was help provided? Was time unlimited?
School	Assessment errors.	Was it timed, mixed, noisy, unfamiliar, or subject-level specific?
Tuition	Fresh practice succeeds.	Was the topic announced? Were prompts or worked examples nearby?
Learner	Says the topic is understood.	Can the learner explain, retrieve and use it on a fresh task?

Once the conditions are visible, the disagreement often becomes a useful profile: the learner may be Stable with notes and Fragile without them; Transfer-Ready in untimed work but not Examination-Ready; or independent at home only after a parent chooses the first task.

Longitudinal profiles: describing change without freezing the past

Longitudinal review is useful when the learner has had a persistent or high-cost weak link. The profile should show movement, not simply accumulate incidents. A simple timeline can record the state, evidence and support level at each meaningful review.

Phase	State	Support	Evidence interpretation
Initial	Blocked	Worked example + targeted prompts	First move unavailable independently.
After repair	Fragile	Strategic prompt	Method appears but does not reliably return.
Consolidation	Stable	Generic prompt or none	Reliable on familiar tasks after delay.
Variation	Transfer-Ready	Independent	Survives changed representation and context.
Paper phase	Examination-Ready	Independent under timed mixed conditions	Knowledge and execution remain reliable in full-paper environment.

The timeline should not be forced into a smooth progression. A learner can move backward temporarily when conditions become harder. The question is whether the changed state accurately describes current evidence, not whether the graph looks encouraging.

Regression versus changed demand

When a learner performs worse at a new stage, ask whether the capability itself weakened or the task demand changed. Primary 5, Secondary 1 and Secondary 3 are common examples. The learner may retain the old skill while needing a new integration skill that was never previously required.

A good profile names the new demand. “Previously Stable on direct fraction comparison; now Fragile when the reference whole changes inside multi-step problems” is more accurate than “fractions got worse”.

Normal variability versus meaningful change

Not every weaker performance requires a profile update. Sleep, illness, unusual difficulty, emotional events and random item variation can move one result. Look for patterns that repeat or for evidence strong enough that the cost of ignoring it is high.

The profile should become more cautious when evidence is noisy. “Current performance varied widely across three mixed sets; no single weak link yet explains the pattern” is a legitimate conclusion.

Capability profiles and accessibility: preserving the target

Accessibility is one of the places where profile discipline matters most. The goal is to make the intended capability observable without letting an unrelated access barrier dominate the result. This requires a clear statement of the target.

If the target is…	A support that may preserve the target	A change that may alter the target
Science causal reasoning	Typed response instead of handwriting, if handwriting is not being assessed.	Providing the causal chain itself.
Mathematical problem solving	Clearer visual layout or enlarged text.	Supplying the equation the learner was meant to formulate.
Reading comprehension	Accessible print format where permitted.	Reading the passage aloud if independent reading is part of the target.
Oral argument	Preparation time or visual prompt, where consistent with task rules.	Supplying the reasons the learner must generate.
Handwriting fluency	Appropriate writing tool or lined guide.	Replacing handwriting with typing when handwriting itself is the target.

The profile should record legitimate access conditions consistently. Changing them between checks can make apparent progress or decline uninterpretable.

Do not use the capability profile to challenge formal accommodations

A tuition capability profile is not a mechanism for questioning clinical or school-approved accommodations. If formal access arrangements apply, the educational profile should work within those conditions unless the relevant authorised professionals change them. The tutor’s job is to understand what the learner can demonstrate with the legitimate support in place.

When a capability profile should not be created

• A one-off low-cost mistake with no repeated pattern and no important decision attached.
• A temporary issue already explained by a known event and likely to resolve without special intervention.
• A situation where the task did not actually give the learner an opportunity to demonstrate the capability.
• A domain outside the tutor’s competence where educational evidence cannot support the conclusion being considered.
• A personal characteristic that does not need to be recorded to make the educational decision.
• A medical, psychological or developmental diagnosis that belongs to qualified professionals rather than tuition inference.
• A profile whose only purpose would be to justify an intervention already chosen in advance.

When a capability profile is especially valuable

• The same symptom has survived several rounds of ordinary practice.
• Different adults describe the learner in contradictory ways.
• A child succeeds with support but independent performance is unclear.
• The same error appears across several topics and may share a dependency.
• A major transition has changed performance and the cause is uncertain.
• Tuition is not improving independent work and the support model needs review.
• A bad examination result could lead to overreaction unless the paper is decomposed.
• A learner may be ready for less support, but adults need evidence before releasing it.

The evidence burden should match the consequence

Small instructional changes can be made from lighter evidence. If a learner repeatedly confuses two vocabulary items, a brief contrast activity is low cost. High-consequence decisions—dropping a subject, declaring a learner unready, changing a long-term pathway, escalating support significantly—deserve a much stronger evidence base.

The profile should therefore state the decision it is supporting. This makes it easier to judge whether the evidence is sufficient.

Capability profiles and subject choice

Subject choice decisions can tempt adults to turn current marks into permanent aptitude claims. A capability profile can help by separating prerequisite readiness, interest, current support dependence, transfer capacity and recent trend. It should still avoid pretending to predict the learner’s entire future from a short sample.

For example, A-Math readiness can be profiled through algebraic control, equation solving, graph interpretation, persistence on unfamiliar problems and response to feedback. The result may justify targeted preparation, not a fixed identity.

Capability profiles and enrichment

Enrichment should also be evidence-led. A high-achieving learner may benefit from deeper or more open work when foundational capabilities are stable and the added challenge creates useful transfer rather than simply larger workload. The profile can identify what kind of challenge is appropriate: abstraction, proof, synthesis, real-world application, speed, creativity or independent project design.

Capability profiles and intervention cost

Every intervention has a cost in time, attention, opportunity and learner dependence. A profile should help estimate whether the likely benefit justifies that cost. If a child’s minor weakness is already improving through normal school practice, adding a large tuition intervention may displace more valuable reading, sleep, CCA or independent practice.

The smallest effective intervention is often the most sustainable because it preserves more of the learner’s ordinary life and makes the causal effect easier to observe.

Capability profiles and stopping tuition

A profile can support stopping or reducing tuition by making independence visible. If the learner has demonstrated Stable or Transfer-Ready performance across delayed fresh work, can identify errors, seeks help precisely and maintains progress with reduced prompts, the system should ask whether routine support is still necessary.

The decision should not be driven by fear that marks might fall without tuition. It should be tested: reduce support, preserve monitoring, collect fresh evidence and be willing to re-enter if a real weak link returns.

Capability profiles and changing tutors

Changing tutors can be appropriate when expertise, communication, accessibility, scheduling or instructional fit is poor. But a change should not erase the learner model. Carry forward the evidence while allowing the new tutor to verify it. The goal is continuity without forcing the new tutor to inherit every old interpretation.

Capability profiles and tutor performance

Sometimes the learner profile reveals a teaching problem rather than a learner problem. If several learners remain dependent on the same prompt, if fresh tasks repeatedly fail after apparently successful lessons, or if explanations produce immediate accuracy but no delayed retention, the instructional route deserves review.

A capability profile is therefore also feedback on teaching. It shows whether the intervention changed what the learner can do independently.

Capability profiles and small-group tuition

In a three-learner group, the same lesson can generate three different profiles. One learner may be Blocked and need a model; another may be Stable and need transfer variation; a third may be Examination-Ready and need selective full-paper calibration. Grouping works when shared instruction can coexist with individual evidence and next actions.

The tutor should not force all three learners into one state simply because they share a table.

Capability profiles and homework

Homework is useful profile evidence only when the support conditions are known. A completed worksheet may reflect independent practice, sibling help, answer checking, online tools or repeated parent prompts. Rather than policing the home, ask what kind of evidence the homework can legitimately provide.

Fresh lesson-entry questions can verify whether the homework learning returned independently.

Capability profiles and revision notes

Beautiful notes can support organisation and retrieval, but note quality is not the same as capability. The profile should ask whether the learner can reconstruct the idea, solve a fresh problem or explain the mechanism without reading the notes. If not, the notes may be a resource rather than evidence of learning.

Capability profiles and memorisation

Memorisation is not inherently weak. Some knowledge must become easily retrievable. The profile asks whether memorised knowledge is the target itself or a prerequisite for later use. Vocabulary definitions may need stable recall; Science marks may additionally require mechanism and application. Formula recall may be necessary; method selection remains a separate capability.

Capability profiles and creativity

Creative tasks can still be profiled without reducing creativity to a checklist. Focus on the capabilities that support the task: generating alternatives, selecting among them, developing a coherent choice, adapting to constraints, revising after feedback and communicating the result. The profile should not claim that creativity is one fixed trait.

Capability profiles and collaboration

Collaboration profiles should distinguish contribution, listening, explanation, coordination and individual reconstruction. A learner who speaks less may still contribute high-quality reasoning; a dominant speaker may prevent peers from demonstrating their capability. Use individual follow-up evidence when the decision concerns an individual learner.

Capability profiles and oral participation

Participation frequency is not equivalent to understanding. A learner may speak often but reason superficially, or speak rarely but produce strong written evidence. If participation matters, profile the specific capability: asking clarifying questions, explaining a reason, responding to challenge, building on a peer’s point or presenting to an audience.

Capability profiles and multilingual learners

When a learner uses more than one language, language proficiency and subject knowledge can interact. A profile should avoid attributing every difficulty to general language ability. Where appropriate, compare conceptual understanding through alternative explanations or representations, then separately profile the language needed to express the target in school English.

Capability profiles and technology use

Digital tools can change cognitive load. Typing may reduce handwriting load; graphing software may change calculation demands; AI may supply hints; calculators may free attention for interpretation. Record the tool because it changes the evidence. Then ask whether the tool is part of the intended real-world condition or a scaffold that should later fade.

Capability profiles and AI-assisted learning

AI-assisted work needs at least two profiles when the distinction matters: assisted capability and later independent capability. The learner may use AI to ask for a hint, compare explanations or generate practice. The value is determined by what the learner can later reconstruct, verify and apply without outsourcing judgement.

A useful AI record can be brief: learner’s first attempt, AI contribution, learner revision, verification method, and later fresh task. This keeps the learner’s contribution visible.

Capability profiles and uncertainty over time

Uncertainty should shrink when evidence accumulates, but it can also grow when new contradictions appear. That is not failure. A profile that becomes more cautious after discovering a condition effect is more accurate than one that preserves a simple story.

For example, a learner initially appears Fragile in algebra; later evidence shows algebra is stable but only slow under time pressure. The profile should move from “algebra retention” to “exam-time fluency/pacing”.

A decision matrix for profile state changes

Current evidence	Keep state	Move forward	Move backward / split
Blocked	First move still unavailable without same level of support.	Useful first move appears with less support and survives fresh task.	Split if different task types reveal different causes.
Fragile	Immediate success but delayed/unsupported return remains inconsistent.	Repeated fresh delayed success with low support.	Return to Blocked if first move repeatedly disappears under comparable conditions.
Stable	Reliable familiar performance; transfer not yet demonstrated.	Changed representation/context succeeds repeatedly.	Return to Fragile if delayed familiar work becomes unreliable.
Transfer-Ready	Changed conditions succeed; paper reliability not yet established.	Mixed timed evidence survives switching and interference.	Split if one transfer dimension remains weak.
Examination-Ready	Repeated paper-level evidence remains reliable.	Special profile may retire; normal monitoring.	Return to narrower state if paper failures trace to a specific capability.

The difference between state change and difficulty change

A learner can remain Stable while moving to harder content. Do not interpret every struggle with advanced material as a state regression. State describes reliability relative to a defined capability and task demand. If the capability itself changed, create or revise the profile boundary before changing the state.

The review question that prevents labels

What new evidence would make us change our mind? If the adults cannot answer that question, the profile may have become a belief rather than a working model. Every capability profile should have a falsification route: evidence that would strengthen, weaken or retire the current interpretation.

FAQ: Is a capability profile the same as a learning profile?

The terms can overlap in ordinary use, but this page uses “capability profile” deliberately to keep the description task-specific and evidence-based. Many “learning profiles” become broad descriptions of preferences or personality. The eduKate version is narrower: what capability, under what conditions, with what support, based on what evidence, and what happens next?

FAQ: Can a parent make one without a tutor?

Yes, as long as it stays modest. A parent can identify a repeated symptom, collect one or two pieces of work, note what help was needed and formulate a testable question. Technical subject diagnosis may still require a teacher or tutor, but the parent can replace global worry with useful evidence.

FAQ: Should the child see the profile?

Usually the learner should understand the relevant part in age-appropriate language, especially the strengths, next move and evidence needed for independence. Avoid sharing technical or speculative language that could become identity. The student version should feel like a route, not a verdict.

FAQ: How long should a profile last?

Only as long as it changes a useful decision. Some profiles last days; others may be monitored across a term. Add a review date or retirement condition. Old profiles should not remain active merely because they were once important.

FAQ: Can the profile contain strengths?

It should. Strength evidence helps choose leverage and prevents deficit-only narratives. Record what is already Stable or Transfer-Ready, especially when it can support the weak link. Do not turn strengths into fixed learning types.

FAQ: What if the child performs differently every time?

Variability is itself evidence. Record the conditions and look for what changes with success: time of day, support, topic cue, task type, delay, language load, fatigue, or another variable. If no pattern emerges, keep the profile uncertain rather than forcing a diagnosis.

FAQ: What if the tutor and school disagree?

Compare tasks and conditions before comparing conclusions. The school may be sampling mixed timed performance while tuition samples focused practice. Both can be accurate. The capability profile should explain the difference and design a bridging retest.

FAQ: Does a low mark mean the learner is Blocked?

No. A low mark can come from Fragile recall, weak transfer, examination timing, answer scope, several small errors or one large dependency. Blocked refers to inability to make a usable first move in the defined capability under the stated conditions.

FAQ: Does a high mark mean Examination-Ready?

Not automatically. One high mark may come from a familiar paper, generous support or favourable topic distribution. Examination-Ready is stronger when performance repeats under mixed timed conditions and the learner can recover from difficulty.

FAQ: Can a capability be Transfer-Ready but not Stable?

The states are a practical route rather than a rigid psychometric scale. Usually transfer depends on enough stability, but evidence can be uneven. A learner may succeed on some changed contexts while still showing delayed fragility. Record the actual pattern rather than forcing a neat label.

FAQ: What if the capability changes because the syllabus changes?

Update the profile boundary. A Primary 6 writing capability and a Secondary 1 argument capability may share components but are not identical. Preserve relevant evidence while recognising the new demand.

FAQ: How does this help decide whether tuition is working?

Tuition is working when the targeted capability changes in the intended direction and the learner increasingly carries the work under fresh, delayed and reduced-support conditions. Attendance, worksheet completion and immediate lesson accuracy are weaker indicators if independent evidence does not improve.

FAQ: How does this help decide when to stop tuition?

A profile can show that the original weak link has been repaired, support has faded, fresh independent work is reliable and the learner can monitor future errors. That creates a testable release route rather than an indefinite commitment.

FAQ: What if a profile makes the child feel judged?

Change the language and involve the learner in the evidence. Focus on the task and current condition. Name strengths. Explain that the profile is temporary and can be disproved by new evidence. If the profile is being used socially as an identity, stop and reset the process.

FAQ: Is this only for struggling students?

No. High-achieving learners can use capability profiles to choose enrichment, detect hidden support dependence, test transfer and avoid mistaking familiar success for flexible mastery. The same principles apply; the next action may be deeper challenge rather than repair.

FAQ: How is this different from simply knowing the child well?

Knowing a learner well is valuable, but familiarity can create strong stories that filter new evidence. The profile makes the current claim explicit, bounded and testable. It complements human knowledge by giving the educational decision a clear evidence trail.

FAQ: What is the most important field?

If only one discipline can be preserved, record support and conditions. Many misinterpretations arise because adults remember the correct output but forget how much help, familiarity or time produced it. The second most important field is the fresh retest.

FAQ: What is the fastest way to improve profile quality?

Replace adjectives with observations. Replace the corrected original task with a fresh retest. Replace “mastered” with the strongest condition actually demonstrated. Replace “needs help” with the level and type of support that changed performance.

The capability-profile standard in one page

1. Name a specific capability, not the whole child.
2. Describe the task that sampled it.
3. Record relevant conditions.
4. Record support precisely enough to interpret success.
5. Write what the learner actually did.
6. Locate the earliest weak link supported by evidence.
7. Use Blocked, Fragile, Stable, Transfer-Ready or Examination-Ready only for that capability.
8. State what remains uncertain.
9. Choose the smallest useful intervention.
10. Reserve fresh evidence.
11. Add delay or variation when needed.
12. Update or retire the profile when evidence changes.
13. Route to the correct subject, runtime, examination, tutor or recovery owner.
14. Return increasing control to the learner.

Why this matters across the whole eduKateSengkang estate

The Atlas is built around routing rather than page count. A symptom page identifies what the family can see. Assessment Evidence clarifies what a task can tell us. The capability profile turns observations into a bounded learner model. Subject owners repair knowledge. Learning Practice & Review creates the next evidence. Learning Has Held tests durability. Examination Craft adds paper conditions. Tutor System coordinates support. Yishun owns recovery when ordinary learning breaks down.

Without the capability-profile layer, those owners can become disconnected. The symptom jumps directly to a subject, the subject adds more work, and nobody states what evidence would show that the route should change. The profile is the hinge that keeps diagnosis, intervention and release connected.

Final principle: describe the learning so the learner can outgrow the description

The best capability profile is designed to become obsolete. It names the current condition accurately enough to choose a useful next move, then actively seeks the evidence that will allow the description to change. It does not protect its own authority. It does not turn temporary support into a permanent identity. It does not require the child to keep proving an old weakness.

Describe the capability. State the conditions. Record the support. Admit uncertainty. Test the next move. Update the model. Retire the label before it becomes one.

Decision workflows: start, continue, change, reduce or stop support

A capability profile becomes most valuable when it changes a real decision. Families often frame tuition decisions as yes/no questions—“Does my child need tuition?”—when the evidence may support a more precise choice: begin targeted support, continue because a defined repair is still working, change because the current intervention is solving the wrong problem, reduce because independence has grown, or stop because special support no longer changes the evidence.

The profile does not make the family decision automatically. It organises the evidence so that the decision can be made for an explicit reason rather than from fear, habit, reputation or attendance alone.

Start support

Starting tuition or another intervention is most defensible when the profile identifies a teachable bottleneck, shows that the current environment is not reliably repairing it, and clarifies what additional support would actually do. “Marks are low” is not enough by itself. The profile should name the capability, state the current conditions, record what help has already been tried and specify the expected evidence of improvement.

A learner who is Blocked in algebraic representation may benefit from targeted modelling and guided practice. A learner who is Stable but not Transfer-Ready may need varied problem solving rather than reteaching. A learner who is Examination-Ready but tired from an overloaded week may not benefit from another class at all.

Continue support

Continuing support makes sense when the intervention is still changing the evidence and the learner has not yet reached the intended handover point. The profile should show more than attendance or worksheet completion. It should show state movement, reduced prompting, stronger delayed retrieval, transfer or more reliable paper performance.

Continuation should still have a review point. “Continue indefinitely because the learner is doing well” can create dependence just as easily as continuing because marks are low.

Change support

Change is justified when the profile shows that the current intervention is not reaching the first weak link, the support level is not fading, or fresh independent evidence remains unchanged despite a fair trial. Change may mean a different method, task type, tutor function, lesson emphasis or schedule—not necessarily a different tutor or centre.

Before changing, check intervention fidelity and cost. A good method run inconsistently may need implementation repair. A method that only works under unrealistic conditions may need replacement even if it is theoretically sound.

Reduce support

Reduction is a positive educational decision when the learner has become more independent. Evidence might include successful fresh tasks between lessons, reduced need for prompts, stable performance after delay, more accurate self-correction or stronger help-seeking. Support can be reduced in frequency, intensity, prompt level or scope.

A learner does not need to be perfect before support reduces. The question is whether the remaining support is still necessary to maintain the target capability and whether ordinary school/home practice can now carry more of the route.

Stop support

Stopping special support is appropriate when the capability is sufficiently stable under the relevant conditions and continued intervention no longer changes the next decision. The profile should state the retirement evidence and what ordinary monitoring will continue. If a different capability later becomes weak, a new profile can be opened without resurrecting the old identity.

Decision	Evidence that supports it	Evidence that should make adults pause
Start	Repeated, bounded weak link; current route insufficient; clear teachable job; defined retest.	One bad score; no task evidence; timetable already unsustainable; support job unclear.
Continue	Fresh evidence improving; support fading; state moving; learner using feedback.	Only worksheet completion improving; prompts remain constant; no fresh retest.
Change	Fairly implemented route produces little independent change; wrong owner or construct identified.	Intervention has barely been run; task demand changed sharply; evidence too thin.
Reduce	Independent performance strong enough under relevant conditions; prompts no longer needed as often.	Success exists only immediately after support; delay/transfer untested.
Stop	Profile can be retired; ordinary learning can carry the capability; review condition defined.	Adults are stopping only from frustration or cost while a high-leverage unresolved weak link remains.

The support-fading decision should be profile-driven

Fading support is not an act of toughness. It is a measurement and learning decision. Remove too early and the learner may return to unproductive failure. Remove too late and the learner never gets a fair chance to show independence.

A profile can define the next fading step explicitly: “move from a strategic prompt to a generic prompt”, “remove the supplied diagram”, “stop checking every answer before submission”, or “reduce parent reminders from nightly to one weekly planning check”. Each change should have a fresh observation attached to it.

The difference between reduced support and abandoned support

Reduced support still preserves a route back to help. The learner knows what to try first, how long to persist, what evidence to bring and how to ask a precise question. Abandoned support simply disappears. The profile should help adults reduce scaffolding while strengthening self-monitoring and help-seeking at the same time.

The profile should name the return condition

Every reduction or stop decision should include a return condition: what future evidence would justify reopening the profile? Examples include the same error recurring across three fresh tasks, paper performance dropping despite stable content, a transition creating a new demand, or the learner again requiring a prompt that had previously faded.

Return conditions prevent two opposite errors: keeping support forever and refusing to restore support because the learner had once been considered “fixed”.

A six-week capability-profile cycle

Week	Primary job	Evidence
0	Baseline and first weak link.	Representative fresh task; support conditions recorded.
1	Teach or repair the bounded weak link.	Guided attempt plus one fresh exit item.
2	Reduce support and retrieve after delay.	Fresh independent or lower-support item.
3	Vary representation/context.	Changed-condition evidence.
4	Mix with neighbouring tasks.	Method-selection or integration evidence.
5	Apply relevant paper conditions if appropriate.	Timed/mixed evidence where the construct requires it.
6	Review, update and decide continue/change/reduce/stop.	State change, uncertainty and retirement/next-owner decision.

Six weeks is not a universal prescription. Some repairs need days; some need months. The example shows the logic: teach, return, vary, integrate, review. The profile should move with evidence rather than force every capability through the same calendar.

Capability profiles in parent–tutor communication

A strong parent–tutor conversation becomes shorter when the profile is precise. Instead of “How is she doing?”, the parent can ask: What capability are we currently repairing? What support does she still need? What fresh evidence changed since the last review? What are we testing next? What would justify reducing tuition?

The tutor, in turn, can ask the parent for evidence that the lesson cannot see: whether homework starts independently, whether the learner can explain the method later, whether reminders are changing, or whether the timetable is becoming unsustainable.

Capability profiles in student–tutor communication

The student should be able to describe the profile in ordinary language. If the profile is so technical that the learner cannot understand the next move, it risks becoming something adults do to the learner rather than with the learner.

A useful student summary has three parts: “This is what already works. This is where the route breaks. This is what you will try next before I help.” The retest then lets the student see whether the profile should change.

Capability profiles in school–tuition translation

Tuition should not create a parallel curriculum language that makes school work harder to interpret. When the school uses a different method, terminology or representation, the profile should identify the invariant capability underneath both routes. The tutor can then help the learner translate rather than insist that one method must replace the other.

Where school requirements are explicit—format, notation, response conventions or examination expectations—the profile should respect them. External support should increase the learner’s ability to operate in school, not require a separate private system.

Capability profiles and family priorities

Families have legitimate priorities beyond marks: sleep, mental space, CCA, transport, siblings, finances and time together. The profile should make trade-offs visible without pretending they are purely academic. A low-leverage weakness may not deserve a high-cost intervention if the learner is otherwise progressing and the family week is already full.

Conversely, a small high-leverage repair may be worth prioritising because it reduces repeated frustration across several subjects. The profile helps distinguish those cases.

Capability profiles and school holidays

School holidays can be used for repair, consolidation, enrichment or rest. The profile should decide which. A Blocked prerequisite that will control next term may justify a short focused repair. A learner who is Stable and exhausted may benefit more from rest plus light retrieval than from an intensive programme.

Holiday work should have an exit point. “Catch up” without a defined profile can expand until the whole holiday becomes school.

Capability profiles and new tutors

When choosing a tutor, the profile can clarify the function needed. A learner may need subject explanation, diagnostic repair, deliberate practice, performance coaching, or simply a small-group environment where thinking becomes visible. “Best tutor” is too broad when the tutor job itself is undefined.

After the first few lessons, compare the tutor’s observations with the existing profile. A good tutor may overturn it. That is a strength, not a failure, if the new evidence is better.

Capability profiles and stopping tuition successfully

A successful tuition relationship should contain its own ending logic. The learner increasingly identifies errors, chooses practice, asks precise questions and uses feedback independently. The profile gives concrete evidence for that handover. Stopping then becomes a planned release rather than a sudden withdrawal or an indefinite continuation.

Frequently asked questions

Is a capability profile the same as a learner profile?

Not necessarily. “Learner profile” can mean many things, including broad personal, demographic, preference or developmental information. This page uses a narrower construct: a capability profile describes educational performance on a bounded capability under stated task, support and condition variables. It deliberately avoids turning broad personal characteristics into instructional destiny.

How is this different from saying a child is weak in a topic?

A topic label tells you where the difficulty appeared. A capability profile tells you what failed inside the topic, under what conditions and what evidence should happen next. “Weak in fractions” might become “Stable in equivalence calculations but Fragile when the reference whole changes in contextual problems.” The second description can guide teaching.

Do I need a formal test to build a profile?

No. A marked school task, a fresh tutor item, a delayed retrieval attempt or a carefully observed explanation can all contribute. Formal tests can help, but the profile should still record what they sampled and how much support was present.

Can parents make capability profiles at home?

Yes, at a simple level. Parents can record the repeated problem, one real piece of work, what the child could do before help, what help changed the attempt and what fresh task will be checked later. Parents should avoid making clinical or permanent conclusions from home observations.

How many questions should be used?

There is no magic number. One question can reveal a possible weak link but rarely supports a broad conclusion. Use enough varied evidence to justify the decision you are making. High-consequence decisions require stronger evidence than a small next-practice decision.

Should the learner see the profile?

Usually the learner should understand the current learning job in age-appropriate language. The full professional record may contain detail the student does not need, but the learner should know what already works, what is being repaired, what support is expected to fade and what fresh evidence will show progress.

Can a capability be in two states at once?

A broad capability may split into sub-capabilities with different states. For example, fraction calculation may be Stable while fraction representation in changed contexts is Fragile. If the state depends heavily on conditions, state those conditions rather than forcing one global label.

What if the learner has one very bad day?

Treat it as evidence, but not automatically as a new identity or permanent state. Check whether fatigue, illness, unusual task difficulty or other conditions changed. Retest if the decision matters.

What if the learner improves immediately after explanation?

Record the assisted improvement, then use a fresh task after some delay. Immediate understanding is encouraging but does not yet establish durable independent capability.

What if the learner performs differently with different tutors?

Compare the support conditions, task difficulty, prompts and relationship cues. The difference may reveal hidden support dependence or differences in opportunity to demonstrate the capability. Do not simply average the performances.

Should accommodations be removed for a fair test?

Not when the accommodation is a legitimate access condition for the target being assessed. Record it. Fair comparison depends on keeping the construct and relevant access conditions clear. Removing an accommodation can make the comparison less valid, not more.

Can AI-generated practice be used as evidence?

Yes, if the task quality is suitable and the learner’s use of AI is recorded. If AI supplied hints, explanations or answers, the performance is assisted evidence. Later independent fresh work is needed before claiming independent capability.

How often should a profile be updated?

Update when new evidence changes the next decision. A newly repaired capability may need frequent review; a Stable maintenance capability may not. Avoid continuous micro-updating that reacts to normal noise.

When should a profile be deleted or retired?

Retire it when special routing no longer changes the learner’s programme. Keep only the minimum history needed for continuity. Old weakness should not remain an active identity after strong evidence has moved on.

What if the profile is wrong?

Change it. A capability profile is a model, not a verdict. Being easy to falsify and update is one of its strengths. A profile that cannot be changed by new evidence has become a label.

Can capability profiles predict examination grades?

They can identify capabilities and conditions relevant to examination performance, but they should not be treated as precise grade forecasts. Examination outcomes depend on the sampled content, paper difficulty, execution, marking and many other factors. The profile’s job is the next learning decision.

Can the same profile be used across English, Mathematics and Science?

The fields can be shared, but the evidence and constructs must remain subject-specific. English answer scope, Mathematics representation and Science mechanism explanation are different capabilities even if all are currently Fragile.

Does a high-achieving student need capability profiles?

Only where the profile changes a decision. High marks may hide transfer fragility or support dependence, but there is no benefit in manufacturing profiles for every strong capability. Use profiling when it clarifies enrichment, transfer, independence or a surprising performance pattern.

Does a struggling student need many profiles?

Usually not at once. Start with the highest-leverage weak link. Repairing one prerequisite may improve several symptoms and reduce the need for multiple separate profiles.

Who owns the final decision: parent, tutor or student?

Different decisions have different responsibilities. Tutors make professional instructional judgements within their role. Parents make family and enrolment decisions. Students increasingly carry learning decisions as they mature. The capability profile provides shared evidence so those roles coordinate rather than compete.

The operating rule for the whole system

A capability profile should always answer seven questions clearly: What capability? Which task? Under what conditions? With what support? What did the learner actually do? What remains uncertain? What happens next? If the document cannot answer those questions, simplify it until it can.

Then ask the final question: What fresh evidence would make us change our mind? That question keeps the profile alive, falsifiable and respectful of the learner’s ability to grow.