Secondary Mathematics Tuition · Turning a marked script into the next piece of Mathematics
Maths feedback is useful only when it changes what the learner can do next. A tick, cross, score, comment, worked correction or tutor explanation is evidence about a previous attempt. The learning begins when that evidence is converted into a more accurate model, a repaired decision, a fresh attempt and a later test of whether the repair survived.
This matters in secondary mathematics tuition because students can spend years “correcting” work without reducing recurring errors. They copy the model answer in another colour, acknowledge the explanation, and move on. The page becomes correct. The capability may remain unchanged.
This guide owns the workflow from received feedback to independent transfer. Detailed error classification belongs to How to Stop Repeating Math Mistakes; practice design belongs to How to Practise Maths Effectively; checking belongs to How to Check Maths Answers. Here the narrow question is: once another person or a marking system tells you something about your Mathematics, what should happen next?
The operating loop is:
RECEIVE → INTERPRET → LOCATE THE FIRST FAILED DECISION → RECONSTRUCT → RETRY → CHANGE THE SURFACE → DELAY → RETEST → UPDATE THE LEARNING PLAN.
1. Feedback is not the comment; feedback is the changed action
“Check your signs.”
“Need more working.”
“Wrong formula.”
“Good.”
“Careless.”
These are signals. Some are useful; some are too broad. None completes the learning process by itself.
A learner has used feedback only when they can translate the signal into a specific next action.
“Check your signs” becomes: identify the exact line where a negative factor was distributed incorrectly, reconstruct the rule, then solve a changed example with a negative outside the bracket.
“Need more working” becomes: identify which transition was not justified or inspectable, then rewrite only enough mathematical state to make the route clear.
“Wrong formula” becomes: identify the misclassified object or missed condition that caused the wrong relationship to be selected.
Feedback is therefore a control input, not a decoration on completed work.
2. The three questions to ask when work is returned
Before looking only at the score, ask:
What did I know?
Where did my route first become unreliable?
What evidence would prove that the weakness is repaired?
The first question protects strengths from being erased by disappointment.
The second locates the earliest useful repair point.
The third prevents correction from ending in passive agreement.
A score is a summary. The script is a map.
3. Separate outcome feedback from process feedback
Outcome feedback tells you whether the answer or final result was accepted.
Process feedback tells you something about the route: representation, method choice, reasoning, algebra, communication, accuracy or checking.
Both matter.
A correct answer can hide an unsafe route. A wrong answer can contain substantial correct reasoning.
For learning, do not collapse everything into right or wrong.
Ask which components of the mathematical process were stable.
4. A mark scheme is not the same as teaching feedback
A mark scheme explains how marks are awarded within an assessment. It may identify acceptable results, intermediate states or reasoning. It is not always designed to teach the concept from first principles.
When using a mark scheme, translate it into learning language.
If a mark is awarded for forming a correct equation, ask what interpretation produced that equation.
If a method mark is lost, locate the missing or invalid relationship.
If an accuracy mark is lost, inspect arithmetic, algebra, calculator input, units or rounding.
Do not simply copy the mark-scheme line and assume the capability is repaired.
5. The first wrong decision matters more than the last wrong number
Suppose a six-line solution ends with the wrong answer.
The last line may contain an arithmetic error, but the decisive failure may have happened at line two when the student formed the wrong equation.
Correcting only the final arithmetic does not fix the model.
Read from the start and find the first point where the solution ceased to follow from the problem or previous valid state.
This is the most efficient repair point because every later error may be downstream.
6. Do not call everything careless
“Careless mistake” is often too vague to guide improvement.
Replace it with a mechanism.
Possible mechanisms include:
misread condition;
copied value incorrectly;
lost negative sign;
used wrong percentage base;
skipped unit conversion;
selected a near-neighbour formula;
rounded too early;
entered calculator expression incorrectly;
answered a different target;
failed to check the result against the context.
A mechanism can be trained. “Careless” rarely can.
7. Correct in a different colour only if the colour means something
Many classrooms require corrections in a different colour. That can be useful if it separates original thinking from revised thinking.
But colour alone is not learning.
Use the correction colour for:
the first wrong decision;
the repaired relationship;
one sentence explaining why the original failed;
the result of a fresh retest.
Do not repaint the whole solution merely to make the page look corrected.
8. Reconstruct before you copy the solution
When feedback includes a full worked answer, cover it after identifying the relevant hint.
Try to reconstruct the route.
If you can only follow while looking, the solution has produced recognition, not independent control.
A useful sequence is:
read the comment;
inspect only enough solution to identify the missing idea;
close the solution;
restart from the original problem;
complete independently;
compare;
then solve a changed problem.
This prevents corrections from becoming transcription.
9. The minimum-help principle
Use the smallest amount of feedback that unlocks a productive next attempt.
Possible help levels:
point to the relevant line;
ask a diagnostic question;
name the missing concept;
show the next representation;
show one intermediate step;
show a parallel example;
show the full solution.
Start as low as possible and escalate only when necessary.
This protects learner responsibility.
10. Feedback should target the decision the learner can change
Comments such as “You are weak at algebra” are too broad.
Better feedback identifies an action:
“When expanding a negative bracket, distribute the negative factor to every term and verify equivalence with a test value.”
“Before reverse percentage, state what represents 100%.”
“Before trigonometry, label sides relative to the chosen angle.”
“Before accepting a calculator result, estimate the scale.”
Specific feedback is easier to turn into practice.
11. Feedback on interpretation errors
If the student misunderstood the question, do not begin by reteaching the computation.
Return to language and representation.
Ask the learner to state:
the target;
known quantities;
conditions;
units;
the relationship implied by the wording.
Then change the surface wording and see whether the relationship is still recognised.
For a full reading system, use How to Read Maths Questions.
12. Feedback on method-selection errors
If the student chose an inefficient or invalid method, ask why that method seemed plausible.
This reveals the cue the learner was using.
Perhaps the presence of a triangle automatically triggered trigonometry even though similar figures were the real structure. Perhaps “percentage” triggered multiplication even though the problem required reverse percentage. Perhaps two equations triggered elimination even though substitution was simpler.
Repair the cue, not only the method.
Then provide a contrast pair where both methods are plausible but only one is appropriate or efficient.
13. Feedback on algebra errors
Locate the exact transformation that broke equivalence.
Ask the learner to explain what operation was applied to each side or each term.
Use a simple substitution check if helpful.
Then retest with changed coefficients or signs.
Do not correct ten later lines if the first invalid algebraic step is already known.
14. Feedback on arithmetic errors
Arithmetic errors may need fluency work, notation discipline, estimation or calculator control.
Identify the mechanism.
A learner who repeatedly writes 7 × 8 = 54 has a different problem from a learner who knows 56 but copies 54 from the calculator display.
One needs fact fluency or attention to arithmetic; the other needs transcription control.
Same wrong number, different repair.
15. Feedback on unit errors
When units are wrong, ask whether the learner lost the unit during calculation or never modelled it.
Rewrite the known quantities with units.
Check conversions.
Predict the output unit before calculation.
Then solve a changed problem using different unit scales.
Units should become part of the mathematical state.
16. Feedback on precision and rounding
Determine whether the learner:
misread the required accuracy;
rounded the wrong digit;
rounded intermediate values too early;
confused decimal places with significant figures;
or reported an exact answer as a decimal.
The correction should target that decision.
A short set mixing exact values, decimal places and significant figures can test whether the distinction transfers.
17. Feedback on diagrams
If a geometry solution fails, inspect the diagram.
Were lengths labelled correctly?
Were right angles, parallel lines or equal sides marked only when known?
Did the learner assume the drawing was to scale?
Did the chosen angle control the side labels?
A repaired diagram can fix the route before any formula is touched.
18. Feedback on graphs
Graph feedback should distinguish plotting accuracy from interpretation.
A learner may plot points correctly but misread scale, gradient, intercept, domain or trend.
Ask the student to move among table, graph, equation and verbal description.
Then retest with a different representation as the starting point.
Transfer across representations is stronger evidence than redrawing the same graph.
19. Feedback on proof and reasoning
When reasoning is incomplete, identify whether the conclusion is unsupported, a theorem condition is missing, a case is omitted, or a statement is merely asserted.
Ask:
What is known?
What must be shown?
Which relationship connects them?
Why is that relationship legal here?
Then reconstruct the proof with reasons visible.
Proof feedback should improve the chain, not merely add more words.
20. Feedback on written working
If working is hard to follow, the problem may be communication, but it may also reveal unstable thinking.
Ask the learner to make each state inspectable:
write the equation;
align equivalent transformations;
label substitutions;
keep units;
separate exact and approximate values.
For the broader system, use How to Show Working in Maths.
21. Feedback on calculator use
If the model is correct but the number is wrong, compare the written expression with the calculator input.
Check brackets, signs, mode, exponent scope, fraction grouping and copied output.
Then estimate before re-entering.
Use the dedicated How to Use a Calculator in Maths guide for the complete workflow.
22. Feedback on formula use
If the wrong formula was chosen, ask which condition was missed.
If the right formula was chosen but substitution failed, map every symbol to the problem.
If rearrangement failed, isolate the algebra.
If the student could not find the formula, classify the mathematical object first.
The companion How to Use a Maths Formula Sheet guide treats these layers in detail.
23. Feedback on unanswered questions
A blank answer is not one error category.
Ask why the learner stopped.
Did they not understand the language?
Could they not recall a relationship?
Did they know several methods and freeze?
Did time run out?
Did they abandon the problem after an unproductive first attempt?
Did anxiety interrupt retrieval?
The recovery path depends on the reason for the blank.
24. Feedback on slow performance
If a solution is correct but too slow, do not automatically drill faster.
Locate the time cost.
Was routine algebra effortful?
Was the method inefficient?
Was the learner checking every line excessively?
Was calculator navigation slow?
Was question interpretation repeated?
Fluency work should target the bottleneck.
For a full speed system, use How to Get Faster at Maths.
25. Feedback on a correct answer
Correct work also deserves analysis.
Ask:
Was the method independent?
Was it efficient enough?
Could the learner explain why it worked?
Would the method survive changed numbers or representation?
Was the result checked?
Positive feedback should identify the capability worth preserving, not simply praise the score.
26. Do not repair everything at once
A page can contain several errors. Choose the highest-leverage repair.
If a weak percentage base causes three downstream errors, repair that first.
If the model is correct and only a final rounding instruction was missed, the repair is small.
Prioritisation prevents corrections from becoming exhausting rewrites.
The learner should leave knowing what one or two changes matter most.
27. The correction ladder
Use increasing levels of correction only as needed:
1. identify the location;
2. ask a question;
3. name the concept;
4. show a representation;
5. show one analogous example;
6. reveal one strategic step;
7. show the full solution;
8. reconstruct independently;
9. solve a changed problem;
10. delayed retest.
The ladder ends with independent evidence, not with the explanation.
28. Feedback should not become answer dependence
If a student receives an immediate hint at the first sign of difficulty, they may become skilled at following rather than initiating.
Allow a productive first attempt.
Ask the learner to represent the problem, state what is known, generate a candidate route and identify the exact point of uncertainty.
Then give bounded help.
This makes feedback responsive to actual need.
29. Ask for feedback precisely
Students can improve the quality of help by asking a precise question.
Instead of “I don’t understand”, try:
“I can form the equation, but I do not see why the sign changes here.”
“I know this is a proportion question, but I cannot identify the 100% base.”
“My method works until the final trigonometric step; can you check my side labels?”
“I got a different answer from the mark scheme. Which is the first invalid line?”
Precise questions preserve more learner ownership.
30. Translate teacher shorthand
Teacher comments are often compressed because marking time is limited.
Translate shorthand into an operational instruction.
“Units” → predict and track units from the model.
“Sign” → locate the transformation where sign control failed.
“Method?” → expose the missing relationship or intermediate state.
“Read Q” → state target and conditions before solving.
“SF” → identify required significant figures and rounding point.
If the shorthand is unclear, ask. Do not invent a meaning.
31. Make feedback auditable
For each repaired error, keep a small record:
Original problem.
First failed decision.
Feedback received.
My interpretation.
Fresh attempt result.
Changed problem result.
Delayed retest date and result.
This turns a comment into a traceable learning event.
32. Do not keep every correction forever
Once a repaired skill has survived several changed and delayed tests, archive the record.
The active error log should remain small enough to guide current practice.
A permanent pile of old failures can become psychologically and operationally heavy.
Keep patterns, not clutter.
33. Feedback timing matters
Feedback is most useful when the learner can still reconstruct the attempt but has enough distance to think rather than defend.
Immediate feedback can be useful for a new procedure or dangerous misconception.
Delayed feedback can be useful when the learner first needs a genuine attempt and later comparison.
There is no single timing rule for every task.
The key is that feedback arrives early enough to prevent entrenched error and late enough to preserve meaningful learner effort.
34. The value of self-feedback before external feedback
Before opening the answer or asking the tutor, ask:
What am I least certain about?
What can I check?
Which line deserves inspection?
What does my estimate say?
Can I use another representation?
This self-check creates a prediction.
External feedback can then be compared against the learner’s own diagnosis, strengthening metacognition.
35. Confidence and feedback
Useful feedback should calibrate confidence rather than simply increase it.
A learner who is uncertain but correct needs evidence about why the route was sound.
A learner who is confident but wrong needs a clear contradiction and a repair.
A learner who is correct only with heavy hints should not interpret the outcome as independent mastery.
Confidence should track evidence.
For the broader system, use How to Build Confidence in Maths.
36. Feedback after a school test
Do not begin with “How many marks did I lose?”
Begin with a script map.
Classify each lost mark by first cause.
Group recurring mechanisms.
Identify one or two high-leverage repairs.
Redo selected questions without the answer visible.
Solve changed versions.
Schedule a delayed mixed retest.
Then update the revision plan.
A test can become a diagnostic instrument rather than a historical document.
37. Feedback after homework
Homework feedback should distinguish practice performance from assisted completion.
Mark which questions were independent, hinted, corrected during work, or completed after seeing a model.
Then retest the supported questions later without help.
This prevents a completed homework sheet from being mistaken for independent capability.
38. Feedback after a mock paper
A mock paper includes knowledge and examination control.
Classify errors not only by topic but by process:
retrieval;
method selection;
time allocation;
navigation;
calculator use;
checking;
communication;
fatigue;
late-paper accuracy.
Then repair in smaller drills before running another full paper.
39. Feedback after tuition
A tuition lesson should leave evidence of what the learner can now do with less support.
At the end of a lesson, ask:
What changed?
Which question can you now do that you could not do at the start?
What still needs a hint?
What will be retested next lesson?
This makes feedback part of progression rather than a stream of comments.
40. Feedback in a three-student group
Small groups can create useful peer feedback if roles are controlled.
Student A solves.
Student B checks the first questionable decision.
Student C asks for the reason or proposes an independent check.
Then roles rotate.
Peers should not simply announce the answer. They should help expose the mathematical state.
This makes explanation visible while preserving individual work.
41. Peer feedback needs a protocol
Use three sentences:
“I agree up to…”
“The first line I question is…”
“I would check it by…”
This prevents feedback from becoming vague judgement or answer swapping.
It also trains students to read solutions structurally.
42. Tutor feedback should fade
Early in learning, the tutor may identify the exact misconception.
Later, the tutor can ask the student to locate it.
Then the student can predict which part is vulnerable before marking.
Eventually, the student can self-check and request help only when uncertainty remains.
Good feedback builds its own redundancy.
43. Parent feedback should focus on process evidence
Parents often see only the score. A more useful conversation asks:
Which error type reduced?
Which topic now survives mixed practice?
Can the child correct without copying?
Can they explain why the original was wrong?
Can they perform after delay?
These questions produce a richer picture than one test percentage.
44. Avoid emotional over-reading of one test
One test is evidence from one set of questions under one set of conditions.
Do not convert it immediately into a fixed identity such as “bad at Maths” or “naturally careless”.
Read the script.
Look for repeat patterns across multiple attempts.
Use changed retests.
Separate temporary performance from durable capability.
This keeps feedback actionable.
45. Feedback and marks are not enemies
Marks matter in assessment. But a mark is most educationally useful when linked to the decision that earned or lost it.
For example, losing two marks on a graph question may represent a scale error rather than weak graph understanding. Losing one method mark repeatedly across algebra questions may reveal a more serious equivalence problem.
The number of marks lost is not always proportional to the depth of the learning issue.
46. Prioritise errors by recurrence and leverage
A useful priority matrix asks:
How often does this error recur?
How many later topics depend on this capability?
How many marks or tasks does it affect?
How easy is it to repair?
A recurring signed-number error has high leverage because it can damage algebra, equations, graphs and coordinate work. A rare formatting slip may be lower priority.
Repair does not have to follow question order.
47. Feedback should generate a changed problem
After correcting the original, alter the surface.
Change numbers.
Reverse the unknown.
Change representation.
Change wording.
Mix with a near neighbour.
Remove the chapter label.
If the learner succeeds only on the original correction, the feedback has not yet transferred.
48. Delayed retesting is essential
Immediate success can reflect short-term memory of the explanation.
Return after a delay.
Use a changed problem.
Remove prompts.
Then see whether the learner independently recognises and executes the repaired relationship.
This is stronger evidence that feedback changed capability.
49. Mix repaired skills back into ordinary work
A repaired topic should not remain in a special “corrections” folder forever.
Return it to mixed practice.
The learner must recognise the relationship without being told that this is the thing they just fixed.
Mixed return is where correction becomes usable Mathematics.
50. Feedback should update the study plan
If feedback reveals a high-leverage gap, revision allocation should change.
If a topic is now stable, reduce maintenance frequency.
If a recurring error survives two repairs, change the intervention rather than repeating the same explanation.
If a performance problem is actually time management, do not spend all revision time reteaching content.
The study plan should respond to evidence.
51. Worked example: algebra sign error
Original line:
−2(x − 5) = −2x − 10.
Feedback: “Sign.”
Interpretation: the negative factor was not distributed correctly to the second term.
Repair: −2(x − 5) = −2x + 10.
Why: multiplication by −2 applies to both x and −5.
Independent check: substitute x = 0. Original expression gives 10; repaired expression gives 10; incorrect expression gives −10.
Changed retest: expand −3(2x − 7).
Delayed retest: include a negative bracket later in a mixed algebra set.
The feedback is complete only after the retest.
52. Worked example: reverse percentage
Original problem gives a final price after a discount and asks for original price.
Student subtracts the percentage from the final amount.
Feedback: “100% base?”
Interpretation: the final amount represents the remaining percentage of the original, not 100%.
Repair: final = multiplier × original, so original = final / multiplier.
Changed retest: use a price after an increase.
Delayed mixed retest: place reverse percentage beside direct percentage so the learner must choose.
53. Worked example: geometry condition
Student uses Pythagoras because the diagram “looks right-angled”.
Feedback: “Where is the right angle given?”
Interpretation: method selection was based on appearance rather than a stated or proven condition.
Repair: identify whether a right angle is marked or derivable. If not, Pythagoras is not yet licensed.
Changed retest: present two similar-looking triangles, only one with a valid right-angle condition.
The learning target is condition discipline, not arithmetic.
54. Worked example: graph scale
Student plots the right coordinates in the wrong vertical positions because the axis increments by 2.
Feedback: “Scale.”
Repair: before plotting, read two adjacent labelled ticks and state the increment.
Changed retest: use a graph with a non-unit scale on both axes.
Delayed retest: integrate graph reading into a function question.
A tiny procedural habit can prevent repeated mark loss.
55. Worked example: correct answer, fragile route
A student obtains the correct solution to an equation by performing different operations on the two sides but happens to land on the right value.
Outcome feedback alone would say “correct”.
Process feedback should say that the balance principle was violated.
Ask the student to explain each transformation.
Then solve a changed equation where the unsafe shortcut fails.
Correctness on one instance is not enough evidence of a valid method.
56. Search language and ownership
Students may search “maths feedback”, “how to correct maths mistakes”, “maths tuition”, “secondary maths tuition”, “how to improve maths”, “careless mistakes maths” or “error log”. These phrases overlap but point to different jobs.
This page owns the workflow after feedback arrives. The error-analysis page owns recurring-error diagnosis. The revision page owns the broader revision cycle. The checking page owns independent answer validation.
Clear ownership prevents one giant article from trying to become every study skill at once.
57. Frequently asked questions
Should I redo every wrong question?
Redo the questions that represent a meaningful gap, and prioritise recurring or high-leverage errors. Some one-off slips may need only a quick correction and check.
When should I look at the model answer?
After a genuine attempt and targeted diagnosis. Use the smallest part of the solution needed to unlock reconstruction, then close it and retry.
How many corrected questions prove I understand?
No fixed number. Look for success on changed problems, later retrieval and mixed-topic use without the original feedback visible.
Should I keep an error log?
Yes if it remains actionable. Record mechanisms, repairs and retests, not every wrong answer forever.
What if I disagree with feedback?
Reconstruct the mathematical argument. Check definitions, conditions and valid transformations. If uncertainty remains, ask the teacher or tutor to identify the exact line and reason. Mathematical feedback should be explainable.
What if the teacher only gives a score?
Use the script, mark scheme where available, your own checking and later discussion to infer the first failure. Ask specific questions rather than requesting a complete re-teaching of the paper.
Can feedback make students dependent?
Yes if help arrives too quickly or always supplies the next move. Good feedback should gradually require more self-diagnosis, reconstruction and independent retesting.
58. A twenty-minute feedback routine
Minutes 1–3: ignore the total score temporarily and scan the script for error clusters.
Minutes 4–7: choose the highest-leverage item and identify the first wrong decision.
Minutes 8–10: interpret the feedback or consult the minimum necessary solution.
Minutes 11–14: redo the original from a clean start.
Minutes 15–17: solve one changed problem.
Minutes 18–19: write one retrieval prompt or error-log line.
Minute 20: schedule a delayed mixed retest.
Short feedback sessions work when they end in fresh Mathematics.
59. The feedback release test
A learner is using feedback maturely when they can:
read beyond the score;
locate the first failed decision;
translate comments into actions;
distinguish interpretation, method, algebra, arithmetic, unit, calculator and communication errors;
use minimum necessary help;
reconstruct without copying;
solve a changed problem;
return after delay;
and update future practice from the evidence.
The strongest test is an unlabelled problem weeks later. If the old error no longer appears and the learner can explain the relevant relationship, the feedback has become capability.
60. Final principle
A marked page belongs to the past. Feedback is valuable only if it improves the future.
Do not correct the paper until it looks right. Correct the decision until the learner can make it right again.
Next routes
Return to BTT Secondary Mathematics Tuition Resources. Use How to Stop Repeating Math Mistakes for deeper error analysis, How to Practise Maths Effectively for practice design, and Secondary Mathematics Tuition for the broader Sec 1–4 route.
Appendix A — The feedback conversion laboratory
Feedback becomes powerful when the learner can convert it into a precise repair without waiting for the teacher or tutor to design every next step. The laboratory below is a set of routines for turning comments, marks, model solutions and self-checks into independent mathematical action.
A1. The comment-to-action drill
Take ten real or invented marking comments: “sign”, “units”, “method”, “read question”, “rounding”, “wrong formula”, “not enough working”, “scale”, “check”, “why?”.
For each comment, write a concrete learner action.
“Sign” might become: locate the first transformation involving a negative quantity, restate the legal operation, and solve a changed example with a negative factor.
“Units” might become: rewrite every known quantity with units, convert before substitution, predict the output unit, then solve a changed problem.
This drill prevents shorthand feedback from remaining shorthand in the learner’s mind.
A2. The first-failure drill
Select five wrong multi-line solutions. Ignore the final answer initially. Read from the beginning and mark the first line that is not justified by the question or preceding valid state.
Then classify the failure: interpretation, model, method, algebra, arithmetic, notation, unit, calculator input, precision or communication.
Finally, hide every later line and repair only the first failure. Recompute the rest from the repaired state.
This teaches causal diagnosis rather than cosmetic correction.
A3. The support-level drill
Take one question the learner cannot solve. Offer help in controlled levels.
Level 1: ask what the target is.
Level 2: ask for a representation.
Level 3: name the relevant topic object.
Level 4: point to the missing relationship.
Level 5: show one analogous example.
Level 6: show one strategic step.
Level 7: show the full solution.
Record the minimum level that allowed the learner to resume independently. Retest later with less support. The objective is to move the minimum necessary help downward over time.
A4. The changed-surface drill
After a correction, change the surface immediately.
Alter numbers, variable letters, diagram orientation, units, wording or unknown direction. Keep the underlying relationship stable.
If the learner succeeds only on the original question, the feedback has repaired memory of the answer rather than control of the idea.
Changed-surface testing is therefore the first transfer check after correction.
A5. The delayed-feedback retest
Return after a delay without showing the original comment.
Use a mixed question in which the learner must recognise the repaired relationship independently.
If the same error returns, reopen the diagnostic record and ask whether the original feedback targeted the true mechanism. A repair that fails twice may require a different representation or earlier prerequisite, not more repetition of the same explanation.
Appendix B — Feedback by evidence source
B1. Teacher written comments
Teacher comments are usually compressed. First interpret the comment in the context of the exact line. Do not generalise “sign” into a belief that all algebra is weak.
Ask what the comment points to, then reconstruct the intended mathematical state. If the shorthand remains ambiguous, ask the teacher a precise question.
B2. Tutor verbal feedback
Verbal feedback can disappear quickly because the page contains no record.
At the end of the exchange, the learner should write one line in their own words: “My first wrong decision was…” and one line: “Next time I will…”
Then complete a fresh attempt. The written micro-record converts speech into an auditable change.
B3. Model solutions
A model solution contains far more information than most learners need at once. Use it selectively.
Compare your solution and the model until the first meaningful divergence. Identify whether the model uses a different valid route or whether your route became invalid. If both routes are valid, compare efficiency and clarity rather than forcing imitation.
B4. Mark schemes
Translate mark allocation into mathematical decisions. If a method mark is associated with forming an equation, ask what interpretation generated that equation. If a result mark is lost only because of final arithmetic, do not reteach the entire concept.
Mark schemes are assessment documents. Convert them into learning evidence.
B5. Automated feedback
Online systems may report correct/incorrect, show a hint or classify an error. Treat automated feedback as evidence with limits.
Check whether the system has access to the learner’s actual reasoning. A wrong final answer does not reveal which line failed. Use the original working to diagnose.
Do not let instant answer checking replace a genuine attempt.
B6. Peer feedback
Peer feedback is useful when it focuses on mathematical reasons rather than authority.
A peer should say: “I agree up to this line; I question this transformation because…” or “I would check it by…”
The learner receiving the feedback should verify the claim. Peers can be wrong. The aim is shared reasoning, not vote-based Mathematics.
B7. Self-feedback
Before external help, ask what you can already detect.
Is the answer in the right range?
Do units work?
Does substitution verify the solution?
Does the graph agree?
Where did certainty drop?
Self-feedback creates a prediction that can later be compared with external feedback, improving calibration.
Appendix C — Feedback by error family
C1. Reading and target errors
Symptoms include answering the wrong quantity, ignoring a condition, confusing “increase by” with “increase to”, or overlooking a required form.
Repair by restating the target in a complete sentence, circling conditions, listing known quantities and representing the relationship before calculation.
Changed retest: keep the Mathematics simple but vary the wording. This isolates reading from computation.
C2. Representation errors
The learner may choose an unhelpful diagram, omit a variable definition, mislabel a table or fail to mark a geometric condition.
Feedback should focus on the representation’s job. Ask what information the representation must make visible.
Retest with a problem where a good representation dramatically simplifies the route.
C3. Method-selection errors
The learner knows several methods but chooses one from a weak cue.
Ask why the method seemed plausible. Then contrast it with a near-neighbour method and identify the decision boundary.
Retest with mixed unlabelled questions. The topic heading must not give away the route.
C4. Conceptual errors
A conceptual error means the underlying object or relationship is misunderstood.
Do not begin with speed drills. Rebuild meaning through examples, non-examples, representations and explanation.
Then return to procedure only after the relationship is coherent.
C5. Algebraic transformation errors
Locate the first broken equivalence. Ask what operation was intended and whether it was applied legally to the whole state.
Use substitution as a spot check when appropriate. Then practise changed coefficients and signs.
C6. Arithmetic errors
Separate fact fluency, place value, sign, copying, calculator entry and rushed execution. They require different repairs.
If arithmetic is the only weakness, protect the mathematical method while training the specific numerical bottleneck.
C7. Unit errors
Make units visible at the start. Predict output units. Convert deliberately. Check dimensional form at the end.
A unit error often reveals that quantities were treated as naked numbers.
C8. Precision errors
Separate exactness, decimal places, significant figures and measurement precision. Identify when rounding occurred and whether it was required.
Retest with mixed reporting instructions so the learner must classify the accuracy demand.
C9. Calculator errors
Compare model, written expression, device input and display interpretation. Do not treat all numerical discrepancies as calculator mistakes.
Use the calculator guide when the device layer itself is unstable.
C10. Communication errors
The Mathematics may be correct but insufficiently inspectable.
Require the learner to expose the equation, substitution, transformation, reason, unit or final interpretation that the assessment expects.
Do not reward extra writing for its own sake. Communication should reveal mathematical state.
Appendix D — Feedback after different kinds of assessment
D1. Short quiz
A short quiz is good for fast diagnosis. Because there are few questions, inspect each error deeply rather than generalising from the score.
Use one changed retest per meaningful weakness. If the quiz targeted recent learning, schedule a later mixed return to see whether the knowledge survives beyond short-term familiarity.
D2. Weighted assessment
Map marks by both topic and mechanism. A low topic score may be caused by one repeated algebra weakness. A high topic score may hide heavy hinting during preparation.
After repair, update the next two weeks of study rather than waiting for the next major examination.
D3. Mid-year or end-of-year examination
Large examinations provide richer evidence but can overwhelm the learner.
Start with clusters. Which three mechanisms caused the greatest loss? Which strengths remained stable under time pressure? Which errors appeared late in the paper?
Repair high-leverage clusters first. Do not redo every question in numerical order.
D4. Prelim paper
A preliminary examination often serves both as assessment and simulation.
Separate content weaknesses from paper-management weaknesses. A question left blank because time expired needs a different intervention from the same question attempted early and misunderstood.
Use section timing, navigation and recovery data alongside topic analysis.
D5. Homework
Homework feedback is difficult to interpret unless support level is known.
Mark questions as independent, hinted, checked during work or copied after solution review. The completion state is not the same as the independence state.
Retest supported questions later without help.
D6. Tuition worksheet
A tuition worksheet should not merely produce more marks. Use it to probe the current learner state.
Record where the tutor intervened. At the end, select one problem for a cold re-attempt with no support.
The difference between supported and independent performance is useful feedback in itself.
D7. Mock examination
After a mock, analyse retrieval, method selection, pacing, calculator control, answer checking and late-paper accuracy.
Then train the weakest process in small drills before running another full paper. Full papers are expensive diagnostic instruments; use the evidence they generate.
Appendix E — The feedback record
E1. Minimal record
Keep the record short:
Question / topic.
First failed decision.
Mechanism.
Feedback.
My repair.
Changed retest.
Delayed retest.
This is enough to track learning without creating administrative burden.
E2. When to promote an error
Promote an error to the active feedback record when it recurs, affects many later topics, costs substantial marks, or reveals a conceptual gap.
Do not promote every isolated slip.
E3. When to retire an error
Retire an item after it survives several changed and delayed mixed problems without special prompting.
Retirement means the learner no longer needs the error in active attention. Archive it if useful, but keep the live system small.
E4. Feedback debt
Feedback debt accumulates when marked work is received faster than it is processed. A student may have five corrected worksheets and no repaired skill.
Prevent this by limiting active repairs. Process important feedback before generating large volumes of new marked work.
More feedback is not automatically better if the learner has no time to convert it.
Appendix F — Feedback conversations
F1. Student to teacher
A strong question contains the attempted reasoning:
“I interpreted this as direct proportion because the ratio looked constant, but the mark scheme uses a linear model with non-zero intercept. Is the intercept the reason direct proportion is invalid?”
This allows the teacher to respond to the exact uncertainty.
F2. Student to tutor
Instead of asking “Can you teach this question?”, show the first point of uncertainty.
“I can draw the diagram and identify the right angle, but I cannot decide which side is adjacent to this angle.”
The tutor can then give smaller, more useful help.
F3. Parent to child
Ask process questions rather than interrogating the score.
“Which mistake repeated?”
“Which question can you now redo without the answer?”
“What will you retest next week?”
This keeps the conversation focused on changeable evidence.
F4. Parent to tutor
A useful update might ask:
What is the first recurring weak link?
How much support is currently required?
What evidence would show independence?
Which current school topics depend on this repair?
This produces a more meaningful progress conversation than “Is my child improving?” alone.
Appendix G — Positive feedback done well
G1. Praise the mathematical action
Instead of “You’re good at Maths”, say what was effective:
“You checked the percentage base before calculating.”
“You rejected the first method because the right-angle condition was missing.”
“You caught the calculator output because your estimate disagreed.”
Specific positive feedback makes successful decisions reproducible.
G2. Distinguish independent from supported success
A correct answer after three hints is progress, but it is different from an independent correct answer.
Name both honestly.
“You completed the method after one representation hint. Next time we will see whether you can generate the diagram yourself.”
This keeps confidence calibrated and the next target clear.
G3. Use success to fade support
When a learner succeeds repeatedly, remove a prompt.
Stop marking the theorem condition for them. Stop reminding them to estimate. Stop choosing the method. Let stable capability carry more of the load.
Positive feedback should lead toward reduced external control.
Appendix H — When feedback seems not to work
H1. The same error returns immediately
The feedback may have been understood verbally but not practised. Require a fresh attempt and changed problem before ending the session.
H2. The same error returns a week later
The repair may not have been retrieved after delay. Add spaced retesting and mixed return.
H3. The same error returns only under time pressure
The knowledge may be present but not sufficiently fluent or accessible. Train the process under gradually tighter time constraints without sacrificing accuracy.
H4. The same error appears across many topics
Look for a shared predecessor: signed numbers, fraction control, algebraic equivalence, unit conversion, reading discipline, calculator grouping or another foundational capability.
Repair the shared dependency rather than every surface instance separately.
H5. Feedback creates argument rather than reflection
Return to the mathematical state. Ask both parties to identify the exact line, definition, condition or counterexample.
Mathematics permits disagreement, but the resolution should rest on reasoning rather than authority or tone.
Appendix I — Feedback and motivation
I1. Keep feedback finite
A page covered in corrections can feel like a verdict. Prioritise the few changes with highest leverage. The learner should leave knowing what to do next.
I2. Preserve evidence of what worked
Do not mark only failures. Identify stable reasoning and accurate habits. This prevents the learner from interpreting a mixed script as total incompetence.
I3. Use progress evidence
Compare the learner’s current error pattern with earlier work. Fewer sign errors, faster method selection or successful delayed retrieval may be meaningful progress even before the headline grade changes.
I4. Avoid false reassurance
Do not call supported success independent mastery. Confidence grows best from accurate evidence.
Appendix J — Feedback by school year
J1. Secondary 1
Feedback should protect the transition into symbolic Mathematics. Focus on notation, signed numbers, algebra meaning, ratio, rate, graph reading and geometric conditions.
Teach students early that a correction ends with a new attempt, not with copying.
J2. Secondary 2
As algebra and graphs become infrastructure, identify recurring dependencies. Feedback should increasingly connect topics rather than treating every mistake as local.
J3. Secondary 3
Route differentiation and Additional Mathematics can increase workload. Prioritise feedback. A student cannot repair everything simultaneously.
Track which errors are broad Secondary Mathematics issues and which belong specifically to Additional Mathematics.
J4. Secondary 4
Feedback must integrate content and examination craft. After every timed paper, distinguish knowledge, recognition, pacing, recovery, checking and late-paper accuracy.
The same mark can arise from very different mechanisms.
Appendix K — Feedback in Additional Mathematics
Additional Mathematics feedback often needs to identify structural algebra errors that propagate through dense solutions.
Find the first broken equivalence, identity misuse, domain issue or incorrect function relation. Avoid rewriting an entire calculus or trigonometry solution if one early algebra decision caused the collapse.
Use exact forms and symbolic checks where possible. Then retest the same dependency in another topic so the learner sees that the repair is portable.
Appendix L — A three-student feedback conference
Each learner brings one error from the week.
Student A presents the original reasoning without revealing the teacher’s comment.
Student B identifies the first questionable line and explains why.
Student C proposes a checking route or changed retest.
Then the original feedback is revealed and compared with the group’s diagnosis.
This trains mathematical reading and helps students see that feedback is reasoning about reasoning.
Appendix M — Tutor feedback fading protocol
Stage 1: tutor names the misconception and models repair.
Stage 2: tutor identifies the location; learner explains the misconception.
Stage 3: tutor asks a diagnostic question; learner locates the issue.
Stage 4: learner self-checks and requests targeted help.
Stage 5: learner independently diagnoses, repairs and retests.
Progress is not only fewer wrong answers. It is a reduction in how much external diagnosis is needed.
Appendix N — Thirty-day feedback reset
Week 1: classify
Take the last three marked pieces of work. Identify first-failure mechanisms and build a small active error record.
Week 2: reconstruct
For each active error, redo the original without visible solutions and solve two changed examples.
Week 3: delay and mix
Retest after several days inside mixed sets. Remove prompts.
Week 4: retire or escalate
Retire stable items. For errors that remain, inspect whether the predecessor was misdiagnosed or whether a different explanation/representation is required.
The active feedback system should end the month smaller, not larger.
Appendix O — A feedback dashboard for the learner
Keep four simple columns:
Stable: skills that survive mixed delayed work.
Watch: skills that are mostly correct but fragile under time or changed representation.
Repair: recurring high-leverage weaknesses.
Retest date: the next evidence point.
Do not turn the dashboard into a score. Its job is routing. It tells the learner what to do next.
Appendix P — The full post-test protocol
1. Record the score, then set it aside.
2. Mark every lost question by first-failure mechanism.
3. Group repeated mechanisms.
4. Identify the two highest-leverage repairs.
5. Reconstruct the original questions without solutions.
6. Solve changed versions.
7. Add one note or error-log update only where needed.
8. Schedule delayed mixed retests.
9. Adjust the next study block.
10. Recheck under time if the original weakness involved pacing.
This protocol turns an assessment from an ending into the beginning of the next learning cycle.
Appendix Q — The feedback release test
Give the learner a previously weak concept inside an unfamiliar mixed problem. Do not mention the old feedback.
Watch whether the learner:
recognises the relevant structure;
avoids the former first-failure mechanism;
checks the vulnerable step;
recovers if a new error appears;
and can explain why the old route would have failed.
If those behaviours appear independently, the feedback has transferred into capability.
Appendix R — The feedback compact
I do not copy a correction until I know what failed.
I do not treat a comment as learning until it changes my next action.
I do not call a repair complete until it survives a changed problem and a delayed return.
I do not keep feedback forever after the capability becomes stable.
That compact keeps feedback quiet, precise and useful.
Appendix S — The feedback fieldbook: twenty-four common learner situations
Feedback often fails because the student does not know what to do with a comment after reading it. The situations below convert recurring feedback problems into small mathematical actions.
S1. “The teacher wrote careless.”
Replace the label with a mechanism. Was the problem misread, a sign lost, a value copied incorrectly, a unit omitted, a calculator entry malformed or a final answer left unchecked? Train the mechanism. Do not train “carefulness” in the abstract.
S2. “I copied the correction and still make the same mistake.”
Copying changed the page, not the capability. Close the correction, reconstruct the original problem independently, then solve a changed version. Schedule a delayed mixed retest.
S3. “I understand the teacher’s solution but cannot produce it.”
Use solution fading. Read only to the first missing strategic step. Close the solution and continue. Repeat until the full route can be generated without visible support.
S4. “The mark scheme uses a different method.”
Do not assume your method is wrong. Check whether your transformations are valid and whether the final result satisfies the problem. Compare efficiency and mark requirements. Multiple methods can coexist if they are mathematically sound and appropriately communicated.
S5. “I lost one mark. Does it matter?”
Depth matters more than mark count. One lost mark caused by a recurring algebra equivalence error may have high leverage. Three lost marks caused by a one-off final copy may be less structurally important. Diagnose mechanism and recurrence.
S6. “I got zero on the question.”
Do not assume nothing was known. Reconstruct the attempt. Did you understand the target, draw the right representation, recall part of the relationship, or make a correct early step? Preserve stable parts and repair the first break.
S7. “I left it blank.”
Identify why. No idea? Time? Panic? Could not read the wording? Could not choose among methods? A blank has no single diagnosis.
S8. “I only make the error in exams.”
The knowledge may be fragile under time or cognitive load. Recreate progressively timed conditions. Reduce avoidable friction through fluency and a stable first-move routine.
S9. “I make the error only in homework.”
Check the environment. Are you distracted, tired, rushing, or using solutions too quickly? Homework feedback should include conditions of work, not only mathematical content.
S10. “My tutor always catches the mistake before I do.”
Delay tutor intervention. Before receiving help, mark the line you are least certain about and perform one independent check. The goal is to shift error detection gradually to the learner.
S11. “I know I am wrong but do not know why.”
That is useful metacognitive evidence. State what seems inconsistent: sign, scale, graph, unit, condition or substitution. Then narrow the check. Uncertainty can be diagnostic when it is localised.
S12. “I disagree with the answer key.”
Rebuild the mathematical argument. Check definitions, assumptions, exact versus approximate forms and whether multiple valid answers exist. If disagreement remains, ask a teacher or tutor about the precise line. Authority does not replace reasoning.
S13. “The teacher says my working is unclear.”
Find the hidden transition. Add the equation, substitution, transformation or reason needed to make the state inspectable. Do not add paragraphs around already-clear arithmetic.
S14. “I keep losing units.”
Write units with known quantities, predict the output unit and track conversion before calculation. Then use a changed problem with different units to prove the repair.
S15. “I round wrongly.”
Classify the instruction: exact, decimal places, significant figures or contextual convention. Keep internal precision until the required stage. Retest with mixed accuracy demands.
S16. “I choose the wrong formula.”
Ask what feature triggered the choice. Repair the recognition cue and the formula’s conditions. Then use a contrast pair containing a tempting near neighbour.
S17. “I choose a correct but very long method.”
First confirm validity. Then compare with a more efficient route and identify what structural recognition made the shorter route possible. Efficiency feedback should follow correctness, not replace it.
S18. “I get the right answer but the tutor still corrects me.”
The route may be unsafe, unjustified or non-general. Ask which step is mathematically fragile. A correct result can occur by coincidence or an invalid shortcut.
S19. “I need a hint for every hard question.”
Change the request. Before asking, state the target, knowns, representation, candidate method and exact uncertainty. This often reduces the hint to one small piece and preserves ownership.
S20. “My score improved but the same mistakes remain.”
The paper may have sampled the vulnerable skill less heavily. Track mechanism recurrence across several assessments. Grade movement and capability movement are related but not identical.
S21. “My score fell but I feel stronger.”
Inspect the paper. Perhaps the test was harder, time control failed, or one new weak topic dominated. Use comparable evidence before concluding that underlying capability declined.
S22. “I have too many corrections to do.”
Prioritise by leverage and recurrence. Repair the few mechanisms that caused many downstream losses. Do not process fifty corrections with equal depth.
S23. “I finished the correction; can I move on?”
Only after a fresh attempt. Ideally add one changed problem and a delayed retest. Completion of correction writing is not completion of learning.
S24. “When can I remove an error from my log?”
When the repaired skill survives changed, delayed and mixed problems without special prompts. Then archive it. Active feedback should get lighter as independence grows.
Appendix T — A twenty-five-question feedback self-audit
1. Do I look beyond the total score?
2. Can I identify the first failed decision?
3. Do I distinguish concept, method and execution errors?
4. Do I know when a comment is too vague to act on?
5. Can I translate shorthand into a next action?
6. Do I attempt before reading a full solution?
7. Do I use the minimum help needed?
8. Can I reconstruct after closing the solution?
9. Do I solve a changed problem?
10. Do I return after a delay?
11. Do I mix repaired skills back into ordinary work?
12. Do I retire errors that are stable?
13. Do I preserve evidence of what I did correctly?
14. Do I avoid calling every error careless?
15. Do I separate independent from hinted success?
16. Do I ask precise questions?
17. Can I disagree mathematically rather than defensively?
18. Do I use units, scale and substitution as self-feedback?
19. Do I update my study plan from assessment evidence?
20. Do I prioritise high-leverage errors?
21. Do I know when the real issue is time rather than content?
22. Do I know when the real issue is a prerequisite rather than the current topic?
23. Is my active error log small enough to guide action?
24. Is external feedback becoming less necessary over time?
25. Can I explain what changed in my Mathematics after a correction?
Use the audit diagnostically. A “no” is not a score deduction. It identifies the next feedback habit to build.
Appendix U — The feedback independence ladder
Level 1 — Result awareness. The learner knows right or wrong.
Level 2 — Location awareness. The learner can identify where the solution went wrong.
Level 3 — Mechanism awareness. The learner can explain why.
Level 4 — Guided repair. The learner can correct with a targeted hint.
Level 5 — Independent reconstruction. The learner can redo the original without visible help.
Level 6 — Changed transfer. The learner can solve a variation.
Level 7 — Delayed stability. The repair survives time.
Level 8 — Mixed recognition. The learner recognises the relationship without a topic cue.
Level 9 — Self-feedback. The learner catches or localises the error before external correction.
Level 10 — Faded support. Tutor or teacher feedback is needed only for genuinely new or difficult uncertainty.
The ladder describes a direction, not a label. Different topics can sit at different levels for the same learner.
Appendix V — A seven-day post-assessment repair plan
Day 1: map the script
Classify lost marks by first-failure mechanism. Preserve examples of strong reasoning.
Day 2: choose two high-leverage repairs
Relearn only the minimum concept or procedure required. Use precise notes.
Day 3: reconstruct originals
Redo without model solutions visible.
Day 4: vary
Change numbers, wording, representation or target.
Day 5: mix
Place the repaired skill beside near neighbours. Remove topic labels.
Day 6: time
If the original failure involved pressure, complete a short timed set.
Day 7: delayed retest
Test again with no reminder of the old error. Retire stable items and keep unresolved ones active.
Appendix W — Feedback under small-group tuition
Three-pax Mathematics tuition allows feedback to remain visible without becoming tutor domination.
A tutor can teach one shared object, then watch three different first attempts. One learner may fail at interpretation, one at algebra, one at checking. The feedback can therefore be individual even when the lesson is common.
The next step is not to give three complete private explanations. It is to assign the minimum repair each learner needs and then bring the group back together for a shared variation or comparison.
This preserves the advantages of peer explanation while maintaining individual diagnostic precision.
Appendix X — Parent progress without grade obsession
A parent can ask for four kinds of evidence:
Independence: how much help is needed now?
Recurrence: which errors are disappearing?
Transfer: can the child solve changed problems?
Delay: does the learning survive a week later?
Grades remain important in an examination system, but these four indicators often reveal capability change earlier than the next major score.
Appendix Y — The final feedback standard
Mature feedback use is quiet. A learner receives a mark or comment, identifies what it means, repairs the smallest necessary piece of Mathematics, proves the repair on a changed problem, returns later, and moves on.
The process does not require drama, self-judgement or endless rewriting. It requires accurate diagnosis and evidence.
The most valuable feedback is the feedback that eventually makes itself unnecessary.
Appendix Z — Five final feedback workshops
Workshop 1: the invisible-mark review
Take a returned piece of work and cover the teacher’s comments and marks. Re-read your own solutions as if you were the checker. Mark the first line you now question in each wrong solution and write the reason.
Then reveal the external feedback. Compare diagnoses. If they agree, self-feedback is strengthening. If they differ, study the difference: did you miss a condition, focus on a later symptom, or misunderstand what the marker was rewarding?
Workshop 2: correction without the original numbers
After learning from a wrong question, close it. Ask the teacher, tutor or study partner to create the same mathematical structure with different numbers and wording. Solve that version first.
Only then return to the original correction. This reverses the usual pattern and makes it harder to rely on memory of the answer.
Workshop 3: explain the old error
Choose an error that has supposedly been repaired. Explain why the old route was tempting, the exact condition or relationship it violated, and the check that would now catch it.
If the learner can only say “I did it wrong before”, the repair may still be shallow. Understanding an error mechanism makes future self-correction more likely.
Workshop 4: feedback under time
Take a repaired skill and place it inside a short timed mixed set. Do not announce which question targets the old weakness. Track whether the learner still uses the new habit when attention is divided across topics and the clock is running.
If the old error returns only under time pressure, the next job is fluency and automaticity rather than another conceptual explanation.
Workshop 5: feedback fading
Repeat a familiar feedback cycle while reducing external support each round.
Round one: tutor identifies the exact line.
Round two: tutor says only that something is wrong.
Round three: learner performs a self-check before receiving any comment.
Round four: learner predicts the vulnerable step before solving.
The goal is for diagnostic work to migrate toward the learner.
Appendix AA — Feedback quality test
Useful feedback should be:
Specific enough to act on. “Watch your signs” becomes useful only when the learner can locate the vulnerable sign operation.
Small enough to use. A long lecture may contain more information than the learner can convert into the next attempt.
Close enough to the cause. Correct the first failed decision, not only the visible final symptom.
Open enough to preserve thinking. Give the smallest prompt that lets the learner resume rather than replacing the route with the tutor’s route immediately.
Testable. The feedback should imply what changed performance would look like.
Temporary. Once the capability becomes stable, the feedback should fade.
Appendix AB — The difference between correction, repair and transfer
Correction means the original work has been made right.
Repair means the underlying knowledge, method or habit has been changed enough to produce a fresh correct attempt.
Transfer means the learner can use that repair when the surface changes, time passes and the topic label disappears.
These are different stages.
A student may complete a correction without repair by copying. They may complete a repair without transfer by repeating a nearly identical question immediately. The full learning loop requires all three.
Appendix AC — Feedback and independence in a small-group Mathematics system
The advantage of a small group is not simply that each learner receives more comments. It is that feedback can be targeted while other learners contribute useful comparison and explanation.
One learner may need a condition prompt, another a representation prompt, and another no hint at all. After those individual repairs, the group can return to one changed problem and compare routes.
The tutor’s long-term objective should be visible: fewer interventions, more precise learner questions, stronger peer reasoning and independent delayed retests.
Appendix AD — Final feedback release standard
Take a skill that produced a recurring error earlier in the term. Place it inside a new mixed problem with different surface features. Do not mention the history.
The learner should recognise the relevant relationship, avoid or catch the former error, complete the solution, check the vulnerable step and explain why the old route would fail.
Then wait. Retest again later.
If the capability survives both changed context and delay without special prompting, the feedback can leave the active system.
Feedback is complete when the learner no longer needs to remember the feedback because the Mathematics itself has changed.
Appendix AE — Commissioning note: what “feedback ready” means
A learner is feedback ready when a mark or comment can be converted into a next mathematical action without requiring somebody else to redesign the entire lesson. The student can find the first failed decision, separate conceptual weakness from execution error, and identify whether the repair belongs to reading, representation, method selection, algebra, arithmetic, units, calculator use, precision, communication or examination control.
The learner can then use the minimum necessary help, reconstruct the solution without copying, and solve a changed problem. They understand that immediate success is not enough: the repaired skill must return after delay and inside mixed work before it can leave the active error system.
Feedback maturity also includes knowing when nothing major needs repair. A correct, independent and well-checked solution should be allowed to remain evidence of stability rather than becoming another opportunity for unnecessary commentary.
The final standard is simple: feedback has done its job when the learner can diagnose more, request less help, repair precisely and prove the change on future Mathematics.