Is This a Mathematics Knowledge Problem or an Examination Problem?

Part I — Diagnose the Broken Link Before Prescribing More Mathematics

A low Mathematics mark can be produced by very different mechanisms. The learner may not understand the concept. They may understand but fail to retrieve it after time. They may remember the method but fail to recognise when it applies. They may choose correctly but execute inaccurately. They may know the Mathematics yet run out of time, overcheck, panic after one difficult item or leave too many accessible marks untouched.

Those problems can produce the same percentage while requiring opposite interventions. A content-heavy response to an examination-control problem wastes time. A paper-strategy response to a real knowledge gap leaves the underlying weakness untouched. Diagnosis therefore begins by locating where the chain first breaks.

Understand → retrieve → recognise → choose → execute → manage time → recover → check → convert knowledge into marks.

Twelve links in the Mathematics performance chain

Understand

Diagnostic test. Can the learner explain the relationship, representation or concept with generous time?

Interpretation. If no, treat the problem as conceptual before worrying about exam technique.

Action. Use linked representations, explanation and simple examples before adding pressure.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Prerequisite

Diagnostic test. Does an older skill such as fractions, signs, ratio or algebra block the current concept?

Interpretation. If yes, repair the minimum necessary dependency and reconnect to current work.

Action. Do not restart the entire earlier syllabus unless evidence demands it.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Retrieve

Diagnostic test. Can the learner bring the relevant method back after a delay without notes?

Interpretation. If no, use active recall and spacing rather than assuming the concept was never learned.

Action. Retest after another delay.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Recognise

Diagnostic test. Can the learner see that the recalled method belongs in an unfamiliar question?

Interpretation. If no, use changed surfaces and mixed recognition tasks.

Action. Topic headings and recent examples may have been hiding the problem.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Choose

Diagnostic test. Can the learner select among two or more plausible methods?

Interpretation. If no, use interleaving, contrast sets and first-move analysis.

Action. Method knowledge and method selection are separate.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Execute

Diagnostic test. Can the learner carry the chosen route accurately?

Interpretation. If no, isolate arithmetic, algebra, notation, units, calculator or working-structure errors.

Action. Do not reteach the concept if selection was already sound.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Transfer

Diagnostic test. Can the same Mathematics survive different wording, diagram orientation or context?

Interpretation. If no, vary surface while preserving structure.

Action. Transfer failure can look like new-topic weakness even when the underlying concept is known.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Manage time

Diagnostic test. Can the learner complete comparable known work under realistic timing?

Interpretation. If no, diagnose reading, selection, fluency, checking and stall time separately.

Action. ‘Too slow’ is not a sufficiently precise diagnosis.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Recover

Diagnostic test. Can the learner protect the rest of the paper after one difficult question?

Interpretation. If no, train leave-return, re-entry and local recovery.

Action. A single stall should not be allowed to become a paper-wide collapse.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Check

Diagnostic test. Can the learner detect likely errors efficiently?

Interpretation. If no, build a personal verification hierarchy.

Action. Checking is a trainable paper skill, not a vague instruction.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Sustain

Diagnostic test. Does performance deteriorate badly late in the paper?

Interpretation. If yes, inspect pacing, fatigue, workload and earlier overinvestment.

Action. Late-paper decline can coexist with strong content knowledge.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Independence

Diagnostic test. Can the learner perform without tutor cues, chapter labels or model examples?

Interpretation. If no, track prompt burden and fade support.

Action. Supported tuition performance should not be mistaken for exam readiness.

The first broken link is usually the best starting point. Later failures may be downstream effects. For example, a learner who cannot retrieve a method may also look slow and anxious because the paper forces a long memory search.

Fifty symptom patterns and what they suggest

Cannot solve even with unlimited time

Observed pattern. The first suspicion is concept or prerequisite rather than examination technique.

Diagnostic move. Use a simple representative problem and ask for explanation without time pressure.

What it may mean. If the idea remains unclear after cues, rebuild understanding before timed practice.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can solve after one small hint

Observed pattern. The knowledge may exist but be inaccessible or recognition may be weak.

Diagnostic move. Record what decision the hint supplied.

What it may mean. Retest later with a weaker cue and changed question.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can solve when told the topic

Observed pattern. Recognition is weaker than topic knowledge.

Diagnostic move. Remove chapter labels and use mixed questions.

What it may mean. If method selection improves, the issue was not raw knowledge.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can solve when shown the formula

Observed pattern. Formula retrieval may be weak.

Diagnostic move. Ask whether the learner understands conditions and can use the formula once recalled.

What it may mean. Use active recall rather than broad reteaching.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can recall formula but chooses wrong formula

Observed pattern. Selection is the problem.

Diagnostic move. Use contrast sets and condition-based reasoning.

What it may mean. Do not spend most practice time memorising formulas already known.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can choose method but makes algebra errors

Observed pattern. Execution is the problem.

Diagnostic move. Classify sign, fraction, substitution or rearrangement errors.

What it may mean. Repair the exact operation and verify in several topics.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can solve untimed but not timed

Observed pattern. Knowledge exists more strongly than paper performance.

Diagnostic move. Measure where time disappears.

What it may mean. Use timed sections only after identifying the bottleneck.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Can solve blanks correctly after exam

Observed pattern. Blank answers may reflect time or recovery rather than knowledge.

Diagnostic move. Compare where the paper first became delayed.

What it may mean. Train pacing and leave-return if appropriate.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Cannot solve blanks even after exam

Observed pattern. The blank likely includes a real knowledge or recognition gap.

Diagnostic move. Use untimed diagnosis before calling it a pacing issue.

What it may mean. Repair the missing link directly.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Homework is excellent, tests are weak

Observed pattern. Homework conditions may be recent, topical, supported or untimed.

Diagnostic move. Compare cold mixed work with homework.

What it may mean. The gap itself is diagnostic.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Tuition work is strong, school papers are weak

Observed pattern. Tutor prompts may be carrying method selection or monitoring.

Diagnostic move. Run silent first attempts and independent timed work.

What it may mean. Track supported-versus-independent performance.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

School work is strong, tuition diagnostic is weak

Observed pattern. The diagnostic may be unfamiliar or the learner may rely on school-specific patterns.

Diagnostic move. Compare multiple evidence sources before concluding broad weakness.

What it may mean. One unfamiliar test should not overrule months of stable work without corroboration.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Scores vary widely

Observed pattern. Knowledge may be inconsistent, but paper difficulty, stalls, checking and confidence can also drive variance.

Diagnostic move. Compare mechanisms across several papers.

What it may mean. Aim to reduce avoidable variance, not only raise the peak.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Scores are consistently low

Observed pattern. A broad knowledge or prerequisite problem becomes more plausible.

Diagnostic move. Still inspect paper control because low knowledge and weak examination skill can coexist.

What it may mean. Prioritise high-leverage foundations first.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Scores are consistently high but below target

Observed pattern. The issue may be should-own losses, efficiency or higher-demand transfer.

Diagnostic move. Inspect recurring small losses and hard-question conversion.

What it may mean. Do not rebuild secure fundamentals unnecessarily.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Many careless errors

Observed pattern. The label is too broad.

Diagnostic move. Classify signs, transcription, units, rounding, reading, calculator and checking.

What it may mean. Use mechanism-specific prevention routines.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Many wrong methods

Observed pattern. Recognition and selection deserve attention.

Diagnostic move. Score first moves separately from final answers.

What it may mean. Use interleaving after individual methods are stable.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Many wrong final answers with right methods

Observed pattern. Execution and checking deserve attention.

Diagnostic move. Locate the first wrong line and classify the mechanism.

What it may mean. Use deliberate accuracy work.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Many unanswered questions

Observed pattern. Separate never reached from reached-but-abandoned.

Diagnostic move. The first suggests pacing; the second may suggest selection or recovery.

What it may mean. Use different interventions for each.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

One question destroys the rest of the paper

Observed pattern. Recovery is a major examination bottleneck.

Diagnostic move. Measure stall duration and downstream accuracy.

What it may mean. Train leave-return rather than reteaching the whole syllabus.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

First page is weak, later paper improves

Observed pattern. Start anxiety, rushing or cold retrieval may be involved.

Diagnostic move. Standardise the opening routine and compare first-page mechanisms.

What it may mean. Do not assume the whole paper reflects weak knowledge.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Final pages are weak

Observed pattern. Pacing, fatigue or earlier overinvestment may be driving deterioration.

Diagnostic move. Compare time checkpoints and late-paper accuracy.

What it may mean. Use full-paper or section-based stamina work.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student changes correct answers to wrong

Observed pattern. Checking calibration is weak.

Diagnostic move. Review why changes were made and what evidence supported them.

What it may mean. Train risk-based checking.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student never changes any answer

Observed pattern. Self-monitoring may be weak.

Diagnostic move. Use planted-error checking tasks.

What it may mean. Build a small personal verification hierarchy.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student uses calculator for simple work

Observed pattern. Fluency or confidence may be weak.

Diagnostic move. Compare mental/symbolic skill with calculator dependence.

What it may mean. Train number sense and efficient tool use rather than banning calculators blindly.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student refuses calculator where useful

Observed pattern. Mechanical work may consume time unnecessarily.

Diagnostic move. Clarify when calculator use is permitted and efficient.

What it may mean. Treat tool choice as a performance decision.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student can explain but not solve

Observed pattern. Conceptual language may be stronger than execution.

Diagnostic move. Observe the exact step where written work breaks.

What it may mean. Repair procedure or notation rather than demanding more explanation.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student can solve but not explain

Observed pattern. Procedural knowledge may be strong but conceptual articulation weak.

Diagnostic move. Use changed problems and non-examples to test depth.

What it may mean. Do not assume lack of verbal fluency equals lack of Mathematics.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student remembers methods only after seeing example

Observed pattern. Recognition may be strong but recall weak.

Diagnostic move. Use delayed active recall before model access.

What it may mean. Test again after a gap.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student remembers methods but not when to use them

Observed pattern. Selection is weak.

Diagnostic move. Use unlabeled mixed questions.

What it may mean. Ask for a structural reason before calculation.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student gets unfamiliar wording wrong

Observed pattern. Transfer or reading may be weak.

Diagnostic move. Hold mathematics constant while changing surface.

What it may mean. If performance recovers after paraphrase, representation may be the key bottleneck.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student gets familiar worksheets right

Observed pattern. Source familiarity may inflate performance.

Diagnostic move. Use unseen sources and changed order.

What it may mean. Compare method selection under novelty.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student is accurate on short questions, inaccurate on long ones

Observed pattern. Coordination, working memory or error propagation may be the issue.

Diagnostic move. Track first-error point and intermediate-state control.

What it may mean. Use multi-step accuracy practice rather than simple drills.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student starts long questions but stalls halfway

Observed pattern. The issue may be state updating or subgoal generation.

Diagnostic move. Use partial-solution continuation and target ladders.

What it may mean. Do not reduce the problem to ‘doesn’t know the topic’.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student cannot start but can continue from a partial solution

Observed pattern. First-move recognition is weaker than continuation.

Diagnostic move. Train problem entry and representation.

What it may mean. The later Mathematics may already be sound.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student can start and continue but never checks

Observed pattern. Paper reliability is limited by verification.

Diagnostic move. Use checking drills and a fixed time budget.

What it may mean. Do not add content if the route is already strong.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student is slow only in word problems

Observed pattern. Reading, representation and model construction may be the bottleneck.

Diagnostic move. Compare time to parse versus calculate.

What it may mean. Train modeling rather than arithmetic speed.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student is slow only in algebra

Observed pattern. Fluency may be the bottleneck.

Diagnostic move. Use short deliberate algebra practice after confirming understanding.

What it may mean. Measure execution time separately from selection.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student is slow only on unfamiliar questions

Observed pattern. Selection and transfer may be the bottleneck.

Diagnostic move. Use first-move timing and changed surfaces.

What it may mean. Avoid global speed drills.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student performs better orally than in writing

Observed pattern. Notation or written organisation may be weak.

Diagnostic move. Compare reasoning with actual paper working.

What it may mean. Train concise written externalisation.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student performs better in writing than orally

Observed pattern. Verbal expression may not reflect mathematical weakness.

Diagnostic move. Use written evidence and transfer tests.

What it may mean. Do not overdiagnose from conversational hesitation.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student improves immediately with tutor beside them

Observed pattern. Reassurance or subtle cues may be affecting performance.

Diagnostic move. Use neutral silent observation.

What it may mean. Track what happens when social support is removed.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student improves after one easy success

Observed pattern. Confidence may be state-dependent.

Diagnostic move. Use representative difficulty next.

What it may mean. Do not mistake one confidence boost for structural repair.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student panics despite strong knowledge

Observed pattern. Examination conditions may be the main limiting factor.

Diagnostic move. Use controlled simulations and recovery protocols.

What it may mean. Keep content on maintenance rather than reteaching everything.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student is calm but consistently wrong

Observed pattern. Confidence is not the issue.

Diagnostic move. Return to concept, prerequisite, recognition or execution diagnosis.

What it may mean. Do not treat calmness as readiness.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student gets marks only after intensive recent revision

Observed pattern. Memory durability is weak.

Diagnostic move. Use spaced retrieval and cold paper checks.

What it may mean. Do not equate freshness with mastery.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student stays strong after long gaps

Observed pattern. Knowledge may be secure enough for maintenance.

Diagnostic move. Reduce dedicated review and target other weaknesses.

What it may mean. Use papers as natural monitoring.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student loses marks mainly to units and rounding

Observed pattern. Knowledge may be broadly secure.

Diagnostic move. Use risk-based execution and checking practice.

What it may mean. This is an exam-conversion problem more than a content problem.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student loses marks mainly to misreading

Observed pattern. Reading discipline and representation deserve attention.

Diagnostic move. Track errors before the first calculation.

What it may mean. Use near-similar prompts with different targets.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student loses marks mainly to skipped subparts

Observed pattern. Paper navigation or reading completeness may be weak.

Diagnostic move. Use part-completion scans and visible subpart tracking.

What it may mean. Do not assume missing knowledge.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Student’s best and worst papers differ dramatically

Observed pattern. Variance analysis matters.

Diagnostic move. Compare paper source, sleep, stalls, checking and recurring mechanisms.

What it may mean. Use several papers before deciding the core diagnosis.

Treat the interpretation as a hypothesis until another task confirms it. Diagnosis should reduce uncertainty through controlled comparisons rather than attach a permanent label after one result.

Part I handoff

The first half of diagnosis is now complete: the performance chain has been separated into distinct links, and common symptom patterns have been mapped to testable hypotheses. Part II will provide controlled diagnostic comparisons, intervention matching, worked cases and the rules for deciding whether the primary problem is knowledge, examination control, or both.

Part II — Controlled Comparisons That Separate Knowledge from Examination Control

The cleanest diagnosis often comes from changing one condition at a time. If performance changes sharply when time, topic labels, notes, source familiarity or tutor prompts are altered, the difference tells us where the system is fragile. The point is not to create more tests. It is to design a small comparison that can rule possibilities in or out.

Forty controlled diagnostic tests

Untimed vs timed

Comparison. Use equivalent questions with generous time and then realistic time.

Interpretation. Strong untimed but weak timed performance points toward retrieval speed, pacing, checking, anxiety or stalls.

Next action. Weakness in both conditions makes knowledge, prerequisite or execution problems more plausible.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Topical vs mixed

Comparison. Compare a chapter-labeled set with an unlabeled set using the same methods.

Interpretation. A large mixed drop suggests recognition or method-selection weakness.

Next action. Similar weakness in both suggests concept or execution problems.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Supported vs independent

Comparison. Compare tutor-guided work with silent first attempts.

Interpretation. A large gap suggests prompt dependence or outsourced monitoring.

Next action. A small gap suggests stronger ownership.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Familiar vs unseen source

Comparison. Use a familiar resource and an appropriate unfamiliar one.

Interpretation. A large gap suggests source dependence or weak transfer.

Next action. Stable performance supports broader structural recognition.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Same-day vs delayed

Comparison. Test immediately after learning, then after a gap.

Interpretation. A large delayed drop suggests weak retention or retrieval.

Next action. Stable delayed performance supports durability.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Notes open vs closed

Comparison. Use equivalent tasks with and without examples or notes.

Interpretation. A large gap suggests cue dependence.

Next action. If both fail, understanding may be weak.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Formula given vs recalled

Comparison. Provide a needed relationship in one condition and require memory in another.

Interpretation. A gap isolates formula retrieval.

Next action. If misuse continues even when formula is given, selection or concept is the issue.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Method named vs unnamed

Comparison. Tell the learner the method in one set, remove the label in another.

Interpretation. A large gap isolates recognition and selection.

Next action. If execution fails after naming the method, procedure needs repair.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Simple numbers vs messy numbers

Comparison. Keep structure constant while increasing arithmetic burden.

Interpretation. A large drop points toward execution or working-memory load.

Next action. Stable structure across messy numbers supports robust method knowledge.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Short vs long question

Comparison. Use the same methods inside one-step and multi-step problems.

Interpretation. A large gap points toward coordination, state updating or error propagation.

Next action. If both fail, the individual method remains weak.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Words vs diagram

Comparison. Present the same relationship in prose and visual form.

Interpretation. A gap suggests representation dependence.

Next action. Stable performance supports deeper transfer.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Graph vs equation

Comparison. Represent the same relation visually and symbolically.

Interpretation. A gap identifies the weaker representation link.

Next action. Train translation, not just one side.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

One context vs another

Comparison. Keep structure but change story or application.

Interpretation. A large gap suggests surface-bound knowledge.

Next action. Stable performance supports structural understanding.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Calculator allowed vs unnecessary

Comparison. Compare tasks where calculator use can hide fluency with tasks requiring symbolic control.

Interpretation. A large gap may expose number sense or calculator dependence.

Next action. Do not confuse tool use with understanding.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Model answer visible vs hidden

Comparison. Compare correction with later reconstruction.

Interpretation. A large gap shows recognition of a solution is weaker than generation.

Next action. Do not count copied correction as repair.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

First attempt vs immediate retry

Comparison. Compare raw performance with a second try after minimal feedback.

Interpretation. Large immediate improvement can indicate recognition, retrieval or monitoring rather than complete concept absence.

Next action. Use delayed retest to see whether it lasts.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before cue vs after cue

Comparison. Give one minimal hint after genuine effort.

Interpretation. If the learner suddenly proceeds, record exactly which decision the cue supplied.

Next action. That decision becomes the next training target.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before explanation vs after explanation

Comparison. Teach only the suspected misconception and retest.

Interpretation. Broad improvement supports a concept diagnosis.

Next action. Little change suggests another bottleneck.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before algebra repair vs after algebra repair

Comparison. Repair one shared algebra mechanism while keeping higher topics constant.

Interpretation. Improvement across topics confirms the prerequisite was high leverage.

Next action. No change suggests the diagnosis needs revision.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before retrieval block vs after retrieval block

Comparison. Use active recall on one fragile method.

Interpretation. Improved cold access supports a memory diagnosis.

Next action. If recognition still fails, add selection work.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before interleaving vs after interleaving

Comparison. Train contrast among competing methods.

Interpretation. Improved first moves support a selection diagnosis.

Next action. If execution remains weak, fluency still needs work.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before pacing work vs after pacing work

Comparison. Train one measured time leak in short sections.

Interpretation. Better completion without accuracy collapse supports the pacing diagnosis.

Next action. If not, knowledge or selection may still dominate.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before checking routine vs after checking routine

Comparison. Use a fixed verification budget before and after training.

Interpretation. Fewer preventable losses support a checking diagnosis.

Next action. No gain suggests the check is poorly targeted.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Before recovery training vs after recovery training

Comparison. Use sections containing one hard question.

Interpretation. Shorter stalls and stronger later performance support a recovery diagnosis.

Next action. If ordinary work is still weak untimed, content remains active too.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Early paper vs late paper

Comparison. Compare accuracy, pace and decision quality by paper third.

Interpretation. Late decline suggests stamina, pacing or fatigue.

Next action. Uniform weakness suggests a broader problem.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

First page vs middle page

Comparison. Compare opening performance with later stabilization.

Interpretation. Early weakness may reflect start anxiety, rushing or cold retrieval.

Next action. Use a stable start routine before global conclusions.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Blank question untimed revisit

Comparison. Return to blanks with generous time and no notes.

Interpretation. If solved, time or recovery becomes more plausible.

Next action. If not, inspect knowledge, recognition and representation.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Wrong question with method supplied

Comparison. Give the correct method but not steps.

Interpretation. Recovery suggests selection was the key weakness.

Next action. Continued failure points to execution or concept.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Wrong question with first step supplied

Comparison. Give only the first move.

Interpretation. If the learner continues, problem entry was the bottleneck.

Next action. Serial hints mean the chain is less secure.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Wrong question with representation supplied

Comparison. Provide the diagram, table or equation but not method.

Interpretation. Improvement suggests modeling or representation weakness.

Next action. Train generation of that representation later.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Poor-sleep vs normal-rest

Comparison. Repeat comparable work under normal rest after a fatigued failure.

Interpretation. Large recovery suggests state and workload mattered.

Next action. Do not overdiagnose abnormal conditions.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

School paper vs tuition paper

Comparison. Compare mechanisms rather than raw scores.

Interpretation. Recurring errors across both are stronger evidence than one percentage gap.

Next action. Source-specific swings suggest transfer or familiarity.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Easy paper vs hard paper

Comparison. Track should-own marks and recurring mechanisms on both.

Interpretation. If ordinary losses persist across difficulty, reliability is real.

Next action. Hard-paper score alone should not define diagnosis.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

One-to-one vs silent independent

Comparison. Compare live interaction with no-intervention work.

Interpretation. A large gap suggests social or prompt dependence.

Next action. Use support fading.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Homework vs cold quiz

Comparison. Compare recent assignment success with later no-note retrieval.

Interpretation. A gap suggests freshness, cueing or weak durability.

Next action. Homework completion alone is not mastery evidence.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Practice set vs paper

Comparison. Compare isolated skill success with integrated paper performance.

Interpretation. A gap suggests selection, timing, checking or recovery.

Next action. Do not automatically reteach content.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Full paper vs selected section

Comparison. Use a section to isolate one suspected performance weakness.

Interpretation. If the problem repeats cleanly, diagnosis is stronger.

Next action. If not, whole-paper interaction may be the cause.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Morning vs afternoon

Comparison. Compare only when time-of-day effects are plausible and repeated.

Interpretation. Stable differences may reflect routine, sleep or energy.

Next action. Do not overread one session.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Confidence rating vs actual result

Comparison. Ask for brief certainty ratings before marking.

Interpretation. Mismatch reveals calibration issues.

Next action. Use sparingly and focus on evidence, not self-judgment.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Heavy workload vs reduced workload

Comparison. Remove redundant practice while preserving key targets.

Interpretation. Improved performance supports overload as part of the problem.

Next action. More work was not the correct intervention.

Keep the comparison as clean as practical. If timing, source, difficulty, prompts and topic all change at once, the result may be too confounded to guide teaching.

Once one link is identified, treat it and repeat an appropriate comparison later. Diagnosis earns credibility when the matched intervention changes the predicted behavior.

Part II handoff

Controlled comparisons reduce guesswork. The next layer should convert those findings into matched interventions: concept repair for concept gaps, retrieval for memory gaps, interleaving for selection gaps, precision work for execution errors, and paper-control training for timing, checking and recovery problems.

Part III — Match the Intervention to the Broken Link

Diagnosis matters because different weak links need different treatments. The student who cannot explain a concept should not receive the same programme as the student who understands it but forgets after a week. The student who selects the right method but loses signs should not receive the same programme as the student who chooses the wrong method. The student who knows the paper but runs out of time needs a different intervention from the student who cannot solve the questions untimed.

Thirty matched interventions

Concept reconstruction

Use when. Use when the learner cannot explain the idea or use it untimed.

Intervention. Teach through linked representations, examples and student explanation.

Guardrail. Keep practice simple enough that conceptual meaning is visible.

Verification. Verify on a changed problem after a delay.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Prerequisite bridge

Use when. Use when an older skill blocks current learning.

Intervention. Repair only the necessary dependency and reconnect immediately to the current topic.

Guardrail. Avoid restarting entire earlier syllabuses.

Verification. Verify inside present work.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Active recall

Use when. Use when the learner understands but cannot bring knowledge back after time.

Intervention. Attempt retrieval before notes and use minimal cues.

Guardrail. Keep targets narrow and high value.

Verification. Verify after increasing delays.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Spaced practice

Use when. Use when memory decays across days or weeks.

Intervention. Schedule returns according to retrieval performance.

Guardrail. Widen intervals after repeated success and shorten after failure.

Verification. Verify in mixed work.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Interleaving

Use when. Use when methods are known but confused in mixed questions.

Intervention. Contrast plausible methods without topic labels.

Guardrail. Start with narrow method pairs before broad mixes.

Verification. Verify in authentic sections.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Representation training

Use when. Use when wording, diagrams, graphs or tables block access.

Intervention. Practise choosing and translating representations deliberately.

Guardrail. Keep mathematical structure constant at first.

Verification. Verify across changed surfaces.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Reading routine

Use when. Use when the first error occurs before calculation.

Intervention. Train target, givens, constraints, command words and units.

Guardrail. Use near-similar prompts that require different actions.

Verification. Verify on unseen wording.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

First-move training

Use when. Use when the learner cannot start but can continue after a hint.

Intervention. Practise target identification, representation and one productive first step.

Guardrail. Keep long calculation out of the diagnostic set.

Verification. Verify on unlabeled questions.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Mid-solution continuation

Use when. Use when the learner starts well but stalls halfway.

Intervention. Train state updating, subgoals and method handovers.

Guardrail. Use partial-solution continuation and missing-middle tasks.

Verification. Verify on unfamiliar multi-step questions.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Execution fluency

Use when. Use when route selection is correct but arithmetic or algebra breaks.

Intervention. Practise the exact operation in short deliberate sets.

Guardrail. Return quickly to full-context problems.

Verification. Verify under mixed and timed conditions.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Sign-control routine

Use when. Use when negative signs recur across topics.

Intervention. Make high-risk transformations explicit and add one sign check cue.

Guardrail. Do not slow every line.

Verification. Verify across several topics.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Fraction-control routine

Use when. Use when fractions cause repeated cross-topic breakdown.

Intervention. Repair the exact operation and exact-form handling.

Guardrail. Avoid full chapter reteaching if the issue is narrow.

Verification. Verify inside algebra and applications.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Working-structure repair

Use when. Use when compressed or disorganised working creates errors.

Intervention. Externalise intermediate values and high-risk transitions.

Guardrail. Reduce unnecessary narration.

Verification. Verify on long questions.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Calculator routine

Use when. Use when keying, mode or rounding causes losses.

Intervention. Standardise calculator setup and entry, and require estimation.

Guardrail. Keep mathematical setup independent of keying.

Verification. Verify in timed work.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Units routine

Use when. Use when applications lose units or conversions.

Intervention. Carry dimensional meaning through setup and checking.

Guardrail. Use units as a route cue.

Verification. Verify across contexts.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Rounding routine

Use when. Use when premature approximation contaminates results.

Intervention. Preserve exact values until the proper stage.

Guardrail. Mark approximate values clearly.

Verification. Verify on multi-stage calculations.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Transfer practice

Use when. Use when familiar worksheets are strong but new surfaces fail.

Intervention. Change wording, context, representation or source while preserving structure.

Guardrail. Increase novelty gradually.

Verification. Verify across multiple sources.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Pacing intervention

Use when. Use when known Mathematics is not completed in time.

Intervention. Measure reading, selection, execution, checking and stalls separately.

Guardrail. Train the largest leak in micro-sections.

Verification. Verify improved completion without accuracy collapse.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Leave-return training

Use when. Use when one difficult question consumes disproportionate time.

Intervention. Practise clean leaving, preserved work and re-entry.

Guardrail. Use realistic timed sections.

Verification. Verify lower downstream damage.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Checking hierarchy

Use when. Use when preventable errors survive or checking is random.

Intervention. Rank personal high-probability checks.

Guardrail. Give a fixed time budget.

Verification. Verify higher marks recovered per minute.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Overchecking reduction

Use when. Use when repeated checking causes incompletion.

Intervention. Limit checking to flagged or high-risk work.

Guardrail. Teach evidence thresholds for changing answers.

Verification. Verify more completion with stable accuracy.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Start-routine training

Use when. Use when first-page errors are unusually high.

Intervention. Standardise reading, settling and opening pace.

Guardrail. Keep routine simple and repeatable.

Verification. Verify across several papers.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Stamina training

Use when. Use when late-paper accuracy drops.

Intervention. Adjust pacing, sleep and simulation density.

Guardrail. Use full papers only when content is broad enough.

Verification. Verify final-third stability.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Prompt fading

Use when. Use when tutor-supported work greatly exceeds independent work.

Intervention. Remove one layer of help at a time.

Guardrail. Protect silent first attempts.

Verification. Verify on no-hint mixed work.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Confidence rebuilding

Use when. Use when repeated failure causes avoidance.

Intervention. Create authentic manageable success after real repair.

Guardrail. Avoid artificially easy work.

Verification. Verify willingness and performance on representative tasks.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Confidence calibration

Use when. Use when certainty is much higher or lower than evidence.

Intervention. Use unfamiliar work and occasional confidence ratings.

Guardrail. Connect discussion to actual first-method accuracy.

Verification. Verify improved self-assessment.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Workload reduction

Use when. Use when fatigue and duplicated practice degrade performance.

Intervention. Remove redundant routine work first.

Guardrail. Protect key homework and high-value targets.

Verification. Verify under recovered conditions.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Homework redesign

Use when. Use when assigned work is not addressing the active weakness.

Intervention. Complete required tasks, then add only a short targeted block.

Guardrail. Avoid a third pile of generic worksheets.

Verification. Verify later independent change.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Paper-cycle redesign

Use when. Use when many papers produce flat scores.

Intervention. Pause full papers and act on recurring mechanisms.

Guardrail. Return to papers only after something has changed.

Verification. Verify mechanism movement.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Format change

Use when. Use when one-to-one, group, online or in-person structure is not supporting the needed independence or pacing.

Intervention. Change the operating environment rather than only the worksheet.

Guardrail. Define what improvement the new format should produce.

Verification. Review after a fixed period.

An intervention should have an exit condition. When the mechanism becomes reliable under delayed and increasingly realistic conditions, reduce the dose and redirect time to the next limiting link.

Twenty-five worked diagnosis cases

Low score, many blanks, strong untimed revisit

Evidence. The learner solves most blanks later.

Likely diagnosis. Primary issue likely includes pacing or recovery.

Plan. Use timed sections and stall analysis while maintaining content.

Confirmation. Confirm on a new timed paper.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Low score, full paper attempted, repeated algebra errors

Evidence. The learner reaches nearly everything but converts poorly.

Likely diagnosis. Execution is a major bottleneck.

Plan. Repair shared algebra and working structure.

Confirmation. Confirm recurrence falls across several topics.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Perfect homework, weak test

Evidence. Homework is recent, topical and supported.

Likely diagnosis. Retrieval, selection or transfer may be weak.

Plan. Use cold mixed questions.

Confirmation. Confirm the homework-test gap shrinks.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Strong tuition, weak school paper

Evidence. Tutor cues are frequent.

Likely diagnosis. Prompt dependence is plausible.

Plan. Protect silent first attempts and run independent sections.

Confirmation. Confirm supported-independent gap shrinks.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Strong untimed, weak timed

Evidence. Knowledge exists but access or paper control is fragile.

Likely diagnosis. Time the stages rather than reteaching everything.

Plan. Train the measured leak.

Confirmation. Confirm faster paper completion.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Weak untimed and timed

Evidence. Time pressure does not explain the failure.

Likely diagnosis. Knowledge, prerequisite or representation remains active.

Plan. Rebuild the earliest broken link.

Confirmation. Confirm untimed competence first.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Knows formula, wrong formula chosen

Evidence. Recall is not the main issue.

Likely diagnosis. Method selection is weak.

Plan. Use contrast and condition-based questions.

Confirmation. Confirm correct selection in mixed work.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Chooses method correctly, wrong final answer

Evidence. Selection is sound.

Likely diagnosis. Execution and checking need attention.

Plan. Repair the first wrong operation.

Confirmation. Confirm lower recurrence.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Starts correctly, stalls halfway

Evidence. Opening recognition is sound.

Likely diagnosis. Continuation and state updating are weak.

Plan. Use target ladders and partial solutions.

Confirmation. Confirm longer independent chains.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Cannot start, finishes after first hint

Evidence. Middle Mathematics is stronger than problem entry.

Likely diagnosis. First-move recognition or representation is weak.

Plan. Train entry only.

Confirmation. Confirm independent starts.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Many ‘careless’ mistakes

Evidence. The label hides several mechanisms.

Likely diagnosis. Classify sign, unit, transcription, reading and rounding errors.

Plan. Build specific prevention routines.

Confirmation. Confirm each error family falls.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

One hard question ruins the paper

Evidence. Content may be adequate elsewhere.

Likely diagnosis. Recovery is a major bottleneck.

Plan. Train leave-return and re-entry.

Confirmation. Confirm later questions stay stable.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Final third collapses

Evidence. Accuracy and speed are good early.

Likely diagnosis. Pacing, stamina or fatigue is implicated.

Plan. Use checkpoints and workload review.

Confirmation. Confirm more even paper performance.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

First page collapses

Evidence. Later work is much stronger.

Likely diagnosis. Opening anxiety, rushing or cold retrieval is plausible.

Plan. Train a stable start routine.

Confirmation. Confirm early-page accuracy improves.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Checks constantly, leaves blanks

Evidence. Verification is consuming too much time.

Likely diagnosis. Overchecking and confidence may be the issue.

Plan. Use a checking budget.

Confirmation. Confirm more completion.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Never checks, finishes early

Evidence. Paper time is available but preventable errors survive.

Likely diagnosis. Verification habit is weak.

Plan. Build a personal check hierarchy.

Confirmation. Confirm checking yield.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Strong familiar resource, weak new source

Evidence. Knowledge may be surface-bound.

Likely diagnosis. Transfer is weak.

Plan. Rotate appropriate sources and representations.

Confirmation. Confirm source sensitivity falls.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Strong immediately after lesson, weak a week later

Evidence. Freshness is carrying success.

Likely diagnosis. Durability is weak.

Plan. Use active recall and spacing.

Confirmation. Confirm cold retrieval after delay.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

High confidence, recurring errors

Evidence. Self-assessment is poorly calibrated.

Likely diagnosis. Use unfamiliar independent work and evidence discussion.

Plan. Confirm confidence aligns better with outcomes.

Confirmation. undefined

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Low confidence, strong independent evidence

Evidence. Self-belief lags competence.

Likely diagnosis. Use objective evidence from changed and timed tasks.

Plan. Keep difficulty representative.

Confirmation. Confirm willingness to attempt improves.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

A-Math weak, Mathematics strong

Evidence. One subject has specific gaps while shared algebra may still matter.

Likely diagnosis. Separate subject-specific and shared mechanisms.

Plan. Coordinate workload.

Confirmation. Confirm A-Math improves without E-Math decay.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Both Mathematics and A-Math weak

Evidence. Two full recovery programmes would overload the learner.

Likely diagnosis. Prioritise shared foundations and accessible marks.

Plan. Use targeted sections.

Confirmation. Confirm score floor rises in both.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Strong topics decay during weak-topic repair

Evidence. Allocation is too one-sided.

Likely diagnosis. Move strong topics to light maintenance, not zero exposure.

Plan. Use mixed retrieval.

Confirmation. Confirm strengths remain stable.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Weak topic never improves despite huge volume

Evidence. Practice method may be mismatched.

Likely diagnosis. Re-diagnose concept, retrieval, selection and execution.

Plan. Change intervention.

Confirmation. Confirm the predicted mechanism moves.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Scores vary with sleep and workload

Evidence. State conditions contribute materially.

Likely diagnosis. Reduce redundant work and protect recovery.

Plan. Retest under normal conditions.

Confirmation. Confirm variance falls.

The diagnosis remains a working model until later independent evidence supports it. If the predicted change does not occur, revise the diagnosis rather than simply increasing the same practice.

Parent and tutor rules for diagnostic discipline

Use several evidence sources

Combine scripts, homework, timed work and observation before making broad claims.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Name uncertainty

A provisional diagnosis with a clear test is stronger than false certainty.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Separate cause from symptom

A low mark is the outcome; the mechanism is the cause being sought.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Prefer controlled comparisons

Change one important condition at a time where practical.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Track first wrong decision

Later errors may be downstream effects.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Track prompt burden

Tutor support can hide weak independence.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Track delayed retrieval

Same-day fluency can hide weak durability.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Track method selection

Topic knowledge can hide poor discrimination.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Track should-own losses

Common preventable marks reveal reliability.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Track stalls and completion

Paper control is separate from content.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Do not overreact to one paper

Use trends and mechanism recurrence.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Do not underreact to recurring small errors

Repeated one-mark losses can become expensive.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Move secure links to maintenance

Do not keep every repaired mechanism active forever.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Use exit criteria

Every intervention should know what success looks like.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

Revise the model when evidence disagrees

Diagnosis is a tool, not an identity.

The purpose is to keep the programme adaptive. Good diagnosis reduces wasted practice because it makes the next intervention more specific and the next verification more meaningful.

A compact diagnostic dashboard

• Current score pattern.
• Untimed versus timed gap.
• Topical versus mixed gap.
• Supported versus independent gap.
• Familiar versus unseen gap.
• Delayed retrieval quality.
• First-method accuracy.
• Execution error families.
• Paper completion and largest stall.
• Checking yield.
• Primary diagnosis.
• Competing alternative diagnosis.
• Next controlled test.

The dashboard should be brief. Its role is to make the current hypothesis and next evidence visible, not to create a permanent clinical file for ordinary Mathematics learning.

Final diagnostic standard

The diagnosis is useful when it predicts which intervention should improve the learner. If concept teaching changes performance, the concept diagnosis gains support. If active recall changes cold access, memory was part of the problem. If interleaving improves first moves, selection mattered. If pacing work improves completion without accuracy loss, paper control mattered.

The final question is therefore not whether the student has a knowledge problem or an examination problem as a permanent category. It is: which link is limiting performance now, and what evidence would show that it has changed?

Part IV — Mixed Diagnoses, Exam-Phase Triage and Closure Rules

Many students do not fit neatly into one category. A learner may have one genuine concept gap, one retrieval weakness and one paper-management habit at the same time. The goal is therefore not to force the child into a single label. It is to rank the current limiting links and decide which ones should be trained together, which should be sequenced, and which can remain on maintenance while higher-cost problems are repaired.

Thirty mixed-diagnosis patterns

Concept gap + timing problem

Pattern. The learner cannot solve some questions untimed and also runs out of time on known work.

Plan. Repair the high-leverage concept first while using short pacing drills only on secure Mathematics.

Guardrail. Do not use full-paper speed work to hide a real knowledge gap.

Exit condition. Close the concept target only after untimed and timed transfer both improve.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Prerequisite gap + confidence collapse

Pattern. Older weakness has caused repeated failure and avoidance.

Plan. Repair the prerequisite at minimum sufficient depth and create authentic changed-question success.

Guardrail. Confidence work should follow evidence, not replace the foundation repair.

Exit condition. Close when current work becomes accessible and willingness to attempt rises.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Retrieval gap + method-selection gap

Pattern. The learner forgets some methods and also confuses those that are remembered.

Plan. Use active recall first on the missing methods, then interleave once they are available.

Guardrail. Do not interleave methods that cannot yet be retrieved independently.

Exit condition. Close when delayed mixed selection becomes reliable.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Selection gap + execution errors

Pattern. The learner sometimes chooses the wrong method and sometimes calculates badly after choosing correctly.

Plan. Track first-method accuracy separately from later execution.

Guardrail. Use contrast sets for selection and narrow fluency for the execution mechanism.

Exit condition. Close only when both the opening route and the written process stabilize.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Transfer gap + reading problem

Pattern. Familiar questions are fine, unfamiliar wording produces poor models.

Plan. Vary wording while holding Mathematics constant and train target-givens-constraints reading.

Guardrail. Do not add harder Mathematics until representation is cleaner.

Exit condition. Close when changed wording no longer alters the underlying model.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Pacing + overchecking

Pattern. The student knows the work but spends too much time verifying early answers.

Plan. Set a checking budget and identify the highest-yield checks.

Guardrail. Recover time before asking the learner to ‘work faster’ everywhere.

Exit condition. Close when completion rises without more preventable errors.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Pacing + one catastrophic stall

Pattern. Most time loss comes from one or two difficult questions.

Plan. Train leave-return and re-entry rather than global speed.

Guardrail. Use timed sections containing deliberate stalls.

Exit condition. Close when later paper completion is protected even if hard items remain hard.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Strong content + weak recovery

Pattern. The learner performs well until one difficult item disrupts the paper.

Plan. Keep content on maintenance and train local recovery.

Guardrail. Do not re-teach secure topics merely because the overall score is inconsistent.

Exit condition. Close when downstream accuracy remains stable after difficulty.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Strong content + weak checking

Pattern. Knowledge and completion are good but preventable errors survive.

Plan. Build a personal risk hierarchy and fixed checking window.

Guardrail. Use checking-only drills on completed work.

Exit condition. Close when checking recovers marks efficiently without overchecking.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Weak content + good exam technique

Pattern. The learner manages time and checks well but cannot access enough Mathematics.

Plan. Prioritise high-leverage concept and prerequisite repair.

Guardrail. Maintain paper routines lightly so they do not decay.

Exit condition. Close when accessible paper coverage expands.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Strong homework + weak independent tests

Pattern. Support conditions are carrying performance.

Plan. Compare notes-open, tutor-guided and cold mixed work.

Guardrail. Fade support deliberately.

Exit condition. Close when homework-to-test gap shrinks.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Strong tests + weak homework compliance

Pattern. The learner may know the Mathematics but have workload or motivation issues.

Plan. Separate academic capability from assignment completion.

Guardrail. Redesign homework dose rather than diagnosing broad weakness.

Exit condition. Close when required work becomes sustainable without unnecessary volume.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

High score + high variance

Pattern. Peak performance is strong but reliability is weak.

Plan. Track should-own losses, paper source, stalls and checking.

Guardrail. Focus on reducing avoidable variance rather than broad reteaching.

Exit condition. Close when ordinary performance approaches the learner’s best level.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Low score + low variance

Pattern. The weakness is stable rather than random.

Plan. Prioritise knowledge and prerequisite analysis first.

Guardrail. Add exam-control work only where evidence shows a separate issue.

Exit condition. Close when the score floor begins to rise for known reasons.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Word-problem weakness + strong calculation

Pattern. The learner can execute operations but cannot build the mathematical model.

Plan. Train representation, variable definition and relationship extraction.

Guardrail. Do not prescribe arithmetic fluency that is already secure.

Exit condition. Close when unfamiliar word problems are modeled independently.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Graph weakness + strong algebra

Pattern. The learner can manipulate equations but struggles with visual meaning.

Plan. Train translation between graph, equation, table and words.

Guardrail. Use algebra as support, not as a substitute for graph interpretation.

Exit condition. Close when representation switching becomes reliable.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

A-Math weakness + shared algebra weakness

Pattern. Advanced topics fail partly because core algebra is unstable.

Plan. Repair shared algebra once and verify in both subjects.

Guardrail. Keep A-Math concept diagnosis separate.

Exit condition. Close shared repair only after transfer into both contexts.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

A-Math concept weakness + strong algebra

Pattern. Execution is available but advanced idea is missing.

Plan. Teach the A-Math concept directly.

Guardrail. Do not overpractice foundation algebra that is already reliable.

Exit condition. Close when changed A-Math applications work independently.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Late-paper decline + heavy workload

Pattern. Fatigue may be interacting with pacing.

Plan. Reduce redundant practice and monitor sleep before prescribing more stamina work.

Guardrail. Then retest under normal conditions.

Exit condition. Close when late-paper accuracy improves without excessive extra volume.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Anxiety + real knowledge gap

Pattern. The learner is nervous and also missing important Mathematics.

Plan. Repair the knowledge while using manageable independent exposure.

Guardrail. Do not treat anxiety as the sole cause or as irrelevant.

Exit condition. Close when both competence and recovery behavior improve.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Anxiety + strong knowledge

Pattern. Mathematics is broadly secure but paper conditions disrupt access.

Plan. Use controlled simulations, start routines and local recovery.

Guardrail. Keep content on maintenance.

Exit condition. Close when paper performance approaches untimed capability.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Confidence + overconfidence

Pattern. The learner feels secure but evidence shows recurring selection or execution errors.

Plan. Use unfamiliar no-hint tasks and mechanism feedback.

Guardrail. Avoid argument about ability; let independent evidence calibrate confidence.

Exit condition. Close when self-assessment aligns better with performance.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Low confidence + strong evidence

Pattern. The learner underestimates genuine capability.

Plan. Use representative tasks to show repaired control.

Guardrail. Do not make practice artificially easy.

Exit condition. Close when willingness to attempt and recovery reflect the evidence.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Poor prelim + broad gaps

Pattern. The paper reveals content and performance weaknesses together.

Plan. Prioritise high-frequency, high-transfer knowledge first while protecting accessible paper routines.

Guardrail. Use the remaining runway realistically.

Exit condition. Close each target individually rather than expecting one global turnaround.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Poor prelim + narrow recurring mechanism

Pattern. A large score loss is driven by one repeated issue.

Plan. Use a focused repair-and-retest cycle.

Guardrail. Avoid broad revision that dilutes the high-value target.

Exit condition. Close when recurrence drops across later papers.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Strong prelim + hidden weak link

Pattern. Overall score is good but one dependency recurs.

Plan. Repair proportionately and keep the rest on maintenance.

Guardrail. Do not create an emergency from a strong system.

Exit condition. Close after changed and delayed verification.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Holiday decay + intact understanding

Pattern. The learner remembers meaning but not fluent access.

Plan. Use diagnostic retrieval before reteaching.

Guardrail. Tighten spacing temporarily.

Exit condition. Close when access returns quickly and remains stable.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Holiday decay + lost understanding

Pattern. The learner cannot reconstruct the concept even with cues.

Plan. Rebuild the concept, then restore spacing.

Guardrail. Do not treat all post-holiday weakness as simple forgetting.

Exit condition. Close only after understanding and retrieval both recover.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Tutor dependence + exam panic

Pattern. Live support and emotional reassurance both prop up performance.

Plan. Use silent first attempts and graded realistic simulations.

Guardrail. Fade support while training recovery.

Exit condition. Close when independent performance becomes stable.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Multiple small issues

Pattern. No single dramatic weakness exists, but many one-mark losses accumulate.

Plan. Use an error hierarchy and should-own mark tracking.

Guardrail. Prioritise the most recurrent low-cost repairs.

Exit condition. Close items as recurrence disappears.

When two diagnoses coexist, train them in a sequence or parallel structure that keeps the evidence interpretable. If both conditions change at once, use later tests to confirm which link actually improved.

Diagnosis changes with the examination runway

Far from exams

Use more diagnostic untimed work and deeper concept repair.

There is time to rebuild foundations and verify them after delays.

Paper strategy should not dominate while large gaps remain.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

About two months out

Increase mixed retrieval and timed sections while continuing high-value repair.

The goal is to see whether knowledge is available without chapter cues.

Full papers become more informative as syllabus access broadens.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

About one month out

Distinguish remaining knowledge gaps from paper-control problems aggressively.

Use papers to expose pacing, checking and recovery while targeted repair continues between them.

Do not let paper volume replace intervention.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

Final two weeks

Prioritise high-frequency risks, should-own marks, retrieval and paper routines.

Broad concept rebuilds need stronger justification.

Secure knowledge should be maintained cheaply.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

Final week

Reduce novelty and avoid large diagnostic experiments.

Use personal risk review and familiar representative work.

Protect sleep and stable routines.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

Day before exam

Do not use the day to resolve every diagnostic uncertainty.

Use only light high-value retrieval if helpful.

The system should already be set.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

After the paper

Do not conduct a large emotional post-mortem if other exams remain.

Record only urgent issues and redirect attention.

Full analysis can wait.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

Between papers

Use evidence from the previous paper only when it can improve the next subject or next Mathematics paper without overload.

Avoid late panic volume.

Preserve recovery time.

Time changes the depth of treatment, but not the need to identify the correct mechanism. Even near the examination, knowing whether the risk is retrieval, selection, execution or pacing determines the most efficient final intervention.

Twenty re-diagnosis rules

Recheck concept after teaching

Test. Use a changed untimed problem without the worked example.

Interpretation. If explanation and application now hold, move downstream to retrieval or transfer.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck prerequisite after bridge repair

Test. Place the old skill inside the current topic.

Interpretation. Foundation drill success alone is insufficient.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck retrieval after spacing

Test. Test cold after a longer gap.

Interpretation. If access fails again, tighten spacing or revisit encoding.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck recognition after surface change

Test. Use unfamiliar wording or representation.

Interpretation. If the method is still recognised, transfer is improving.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck selection after interleaving

Test. Use mixed no-label questions.

Interpretation. Score the first method separately from execution.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck execution after fluency work

Test. Use the same operation inside different topics.

Interpretation. Shared mechanism repair should travel.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck timing after micro-section work

Test. Use comparable timed sections.

Interpretation. Completion should improve without a major accuracy trade-off.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck checking after hierarchy training

Test. Use a fixed time budget.

Interpretation. Measure preventable marks recovered.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck recovery after stall drills

Test. Use another difficult timed question.

Interpretation. Downstream performance should be better protected.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck independence after prompt fading

Test. Use silent first attempts.

Interpretation. Support needs should fall on familiar material.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck confidence after authentic success

Test. Use representative unseen work.

Interpretation. Confidence should align with actual control.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck workload after volume reduction

Test. Compare attention, retrieval and paper quality.

Interpretation. If performance improves, overload was part of the system.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck maintenance topics

Test. Use natural mixed work rather than dedicated large sets.

Interpretation. Reactivate only if real decay appears.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck diagnosis after plateau

Test. Change the test condition.

Interpretation. A stagnant intervention is evidence that the model may be wrong.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck after score jump

Test. Confirm on another comparable unseen task.

Interpretation. One peak result should not close every target.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck after score fall

Test. Compare recurring mechanisms before reopening secure areas.

Interpretation. One outlier should not erase broader evidence.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck source sensitivity

Test. Rotate appropriate paper sources.

Interpretation. Stable performance across styles supports transfer.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck paper completion

Test. Track not just finish time but where stalls moved.

Interpretation. Improvement should reflect better allocation, not rushed errors.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck first-method latency

Test. Use short mixed entry drills.

Interpretation. Faster choice matters only when accuracy remains good.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Recheck final diagnosis

Test. Ask which link now limits performance after previous repair.

Interpretation. A successful intervention often reveals the next bottleneck.

The learning system should remain willing to revise its model. Diagnosis is not a one-time label; it is an evolving explanation of what currently limits performance.

Closure criteria: when the diagnosis can move to maintenance

A knowledge target can move to maintenance when the learner can explain it, retrieve it after a gap, recognise it under changed surfaces and use it independently in mixed work. An examination-control target can move to maintenance when the learner performs the paper behavior—pacing, checking, recovery or navigation—reliably across several realistic attempts.

The page’s central principle is therefore practical rather than categorical: find the earliest broken link, choose the intervention that should change it, test whether the predicted change occurs, and then move attention downstream. That is how diagnosis turns a low mark into a useful learning decision instead of a vague demand to study more.

Final diagnostic cross-checks

Cross-check — Concept gap + timing problem

Use a fresh task where the learner cannot solve some questions untimed and also runs out of time on known work. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close the concept target only after untimed and timed transfer both improve., reduce the intervention and move the target toward maintenance. If not, repair the high-leverage concept first while using short pacing drills only on secure mathematics. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Prerequisite gap + confidence collapse

Use a fresh task where older weakness has caused repeated failure and avoidance. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when current work becomes accessible and willingness to attempt rises., reduce the intervention and move the target toward maintenance. If not, repair the prerequisite at minimum sufficient depth and create authentic changed-question success. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Retrieval gap + method-selection gap

Use a fresh task where the learner forgets some methods and also confuses those that are remembered. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when delayed mixed selection becomes reliable., reduce the intervention and move the target toward maintenance. If not, use active recall first on the missing methods, then interleave once they are available. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Selection gap + execution errors

Use a fresh task where the learner sometimes chooses the wrong method and sometimes calculates badly after choosing correctly. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close only when both the opening route and the written process stabilize., reduce the intervention and move the target toward maintenance. If not, track first-method accuracy separately from later execution. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Transfer gap + reading problem

Use a fresh task where familiar questions are fine, unfamiliar wording produces poor models. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when changed wording no longer alters the underlying model., reduce the intervention and move the target toward maintenance. If not, vary wording while holding mathematics constant and train target-givens-constraints reading. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Pacing + overchecking

Use a fresh task where the student knows the work but spends too much time verifying early answers. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when completion rises without more preventable errors., reduce the intervention and move the target toward maintenance. If not, set a checking budget and identify the highest-yield checks. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Pacing + one catastrophic stall

Use a fresh task where most time loss comes from one or two difficult questions. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when later paper completion is protected even if hard items remain hard., reduce the intervention and move the target toward maintenance. If not, train leave-return and re-entry rather than global speed. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Strong content + weak recovery

Use a fresh task where the learner performs well until one difficult item disrupts the paper. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when downstream accuracy remains stable after difficulty., reduce the intervention and move the target toward maintenance. If not, keep content on maintenance and train local recovery. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Strong content + weak checking

Use a fresh task where knowledge and completion are good but preventable errors survive. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when checking recovers marks efficiently without overchecking., reduce the intervention and move the target toward maintenance. If not, build a personal risk hierarchy and fixed checking window. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Weak content + good exam technique

Use a fresh task where the learner manages time and checks well but cannot access enough mathematics. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when accessible paper coverage expands., reduce the intervention and move the target toward maintenance. If not, prioritise high-leverage concept and prerequisite repair. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Strong homework + weak independent tests

Use a fresh task where support conditions are carrying performance. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when homework-to-test gap shrinks., reduce the intervention and move the target toward maintenance. If not, compare notes-open, tutor-guided and cold mixed work. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Strong tests + weak homework compliance

Use a fresh task where the learner may know the mathematics but have workload or motivation issues. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when required work becomes sustainable without unnecessary volume., reduce the intervention and move the target toward maintenance. If not, separate academic capability from assignment completion. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — High score + high variance

Use a fresh task where peak performance is strong but reliability is weak. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when ordinary performance approaches the learner’s best level., reduce the intervention and move the target toward maintenance. If not, track should-own losses, paper source, stalls and checking. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Low score + low variance

Use a fresh task where the weakness is stable rather than random. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when the score floor begins to rise for known reasons., reduce the intervention and move the target toward maintenance. If not, prioritise knowledge and prerequisite analysis first. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Word-problem weakness + strong calculation

Use a fresh task where the learner can execute operations but cannot build the mathematical model. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when unfamiliar word problems are modeled independently., reduce the intervention and move the target toward maintenance. If not, train representation, variable definition and relationship extraction. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — Graph weakness + strong algebra

Use a fresh task where the learner can manipulate equations but struggles with visual meaning. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when representation switching becomes reliable., reduce the intervention and move the target toward maintenance. If not, train translation between graph, equation, table and words. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — A-Math weakness + shared algebra weakness

Use a fresh task where advanced topics fail partly because core algebra is unstable. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close shared repair only after transfer into both contexts., reduce the intervention and move the target toward maintenance. If not, repair shared algebra once and verify in both subjects. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Cross-check — A-Math concept weakness + strong algebra

Use a fresh task where execution is available but advanced idea is missing. could appear. Keep other conditions as stable as practical so the learner’s response can test whether the working diagnosis still predicts performance.

If close when changed a-math applications work independently., reduce the intervention and move the target toward maintenance. If not, teach the a-math concept directly. and repeat the test after a delay.

The purpose of the cross-check is to stop the programme from repeating a diagnosis simply because it was made earlier. Evidence should be allowed to update the plan.

Final Diagnostic Verification Appendix

Can the learner explain the concept untimed?

If no, keep concept repair active before blaming examination technique.

If yes, move downstream and test retrieval, recognition and execution.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner retrieve after a gap?

If no, memory durability remains active.

If yes, do not continue intensive review of material that is already available.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner recognise the method without topic labels?

If no, use interleaving and changed surfaces.

If yes, selection may be more secure than the score suggests.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner choose among plausible methods?

If no, selection is a distinct training target.

If yes, inspect execution rather than reteaching the method family.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner execute accurately untimed?

If no, precision or prerequisite control is weak.

If yes, introduce realistic timing before concluding the paper is solved.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner execute under time?

If no, isolate the time leak.

If yes, do not keep speed work active merely because the student once ran out of time.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner recover from one hard question?

If no, train leave-return and downstream protection.

If yes, one difficult item should no longer control the whole paper.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner check high-risk work efficiently?

If no, verification remains active.

If yes, move checking to normal paper maintenance.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Can the learner perform without tutor prompts?

If no, support dependence remains part of the diagnosis.

If yes, independent evidence should carry more weight than lesson fluency.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

Does the matched intervention change the predicted behavior?

If no, revise the diagnosis.

If yes, the model has earned credibility and the repaired link can move toward maintenance.

Use a changed task rather than the original diagnostic item. The purpose is to confirm the mechanism, not the memory of the test. When the learner passes the relevant check repeatedly, reduce the intervention and identify the next link that limits performance.

A strong diagnosis makes the programme simpler over time: secure knowledge moves to maintenance, fragile links stay visible, and practice is concentrated where it still changes independent performance.

The final standard is prediction. If the diagnosis correctly predicts what kind of practice will improve the student—and that improvement survives delay, changed surfaces and realistic conditions—the diagnosis is useful. If not, change the model rather than asking for more of the same work.