There is a moment in Mathematics that interests me almost as much as the moment before a student begins.
It happens later.
The working is already on the page. Perhaps half a page, perhaps two. The student reaches the answer, looks at it, and something does not sit comfortably.
Maybe a length has become negative.
Maybe a probability is greater than one.
Maybe the graph should be rising but the calculated gradient says otherwise.
Sometimes there is no obvious impossibility. The student simply has the faint suspicion that somewhere, somehow, the Mathematics has stopped telling the truth.
What happens next tells me a great deal.
Some students erase everything.
Some begin again from the first line.
Some stare at the final answer as though enough concentration might persuade it to change.
Some immediately ask, “Where did I go wrong?”
And every now and then, a student does something much more useful.
She goes looking.
That distinction—between wanting the wrong answer removed and wanting to understand where the reasoning first became wrong—is one of the quieter signs of mathematical maturity.
A wrong answer is not one thing
When parents see a wrong answer on a test paper, it is natural to see one event.
Question 7: wrong.
Two marks lost.
Move on.
From a teaching perspective, however, the final cross is only the end of a chain.
A student can arrive at the same wrong answer through very different failures.
The concept may never have been understood.
The right concept may have been recognised, but the wrong method chosen.
The method may be perfectly appropriate and the algebra may have broken halfway through.
The algebra may be correct, but a value may have been copied inaccurately.
The entire solution may be mathematically sound until the final line, where the student answers the wrong quantity.
These are not interchangeable problems.
And this is why I am less interested in the fact that a student is wrong than in what the student does after becoming aware of it.
The repair behaviour often reveals whether the student sees Mathematics as a sequence of answers or as a structure that can be inspected.
My short answer
When a student discovers an error, I want her to locate the earliest point at which the Mathematics stopped being true.
Not merely the line where the calculator produced the wrong number.
Not merely the line the teacher has marked.
The earliest broken relationship.
That is usually where the useful learning begins.
Why starting again can hide the problem
One common response to an error is to erase everything and redo the question from scratch.
It feels responsible.
The student is not giving up. She is trying again.
And sometimes starting again is exactly right. If the page has become so tangled that the reasoning can no longer be read reliably, a clean restart may be efficient.
But I am cautious when restarting becomes the default response.
A complete restart can accidentally destroy evidence.
Suppose a student solves 3(x − 2) = 12 and writes 3x − 2 = 12.
Everything after that line may be executed beautifully.
The student may subtract, divide, simplify and arrive confidently at an answer.
If she sees that the answer does not work and immediately erases the entire solution, she loses the most useful information on the page.
The important fact was not that the answer was wrong.
The important fact was that the distributive law stopped being applied correctly at one particular moment.
If we preserve the page long enough to see that, the correction can become specific.
The problem is no longer “I am bad at algebra.”
It is no longer even “I got this question wrong.”
It becomes: “I distributed the 3 to the first term but not to the second.”
That is a much more repairable problem.
Precision is kinder than vagueness
Students often describe errors in large, discouraging language.
“I always make careless mistakes.”
“I don’t understand A-Math.”
“My algebra is bad.”
“I panic in tests.”
Parents hear these sentences too.
The difficulty is that broad labels rarely tell us what to change.
“Careless” can mean a dozen different things.
Was a negative sign dropped when moving between lines?
Was the question copied incorrectly?
Was a calculator set in the wrong mode?
Was a square omitted?
Was the final answer not checked against the units?
Was the student rushing because time management had already broken twenty minutes earlier?
These failures look similar when reduced to marks.
They are not similar when we try to repair them.
One of the things I want Mathematics tuition to do is make the student’s description of difficulty increasingly precise.
Instead of: “I keep making mistakes.”
I would rather hear: “I usually choose the right method, but when there are several negative signs in the algebra I lose one during expansion.”
That sentence may sound more complicated.
Actually, it is good news.
The student now knows where to look.
Mathematics leaves evidence
There is something unusually helpful about Mathematics in this respect.
Thinking leaves a trail.
Not all of it, of course. Much of mathematical thought happens internally. But a well-written solution leaves enough of a trail that we can often reconstruct what the student believed at each stage.
This is why working is more than something written to satisfy an examiner.
It is diagnostic material.
Consider differentiation.
A student may correctly differentiate y = x³ − 4x² + 2x to obtain dy/dx = 3x² − 8x + 2.
If the question asks for the gradient at x = 2, the student then substitutes 2.
Suppose she obtains 2 when the correct value is −2.
There are several possibilities.
The derivative itself could be wrong.
The substitution could be wrong.
The arithmetic could be wrong.
The negative sign could have disappeared.
If the working is visible, we can isolate the failure.
If the page contains only a final number, almost all of that evidence has vanished.
So when students tell me they prefer doing more steps mentally because it is faster, I understand the attraction.
But there is a point at which compressed working becomes expensive.
If the answer is wrong, there is nothing left to inspect.
Good working is recoverable
I often think of a strong mathematical solution as something that should be recoverable.
Not merely elegant.
Not merely short.
Recoverable.
If something goes wrong halfway through, can the student return to the page and find where the reasoning diverged?
This becomes increasingly important as Mathematics grows more complicated.
In lower Secondary work, an error may occur within three or four lines.
In Additional Mathematics, a question can involve differentiation, substitution, simultaneous relationships and algebraic simplification before the final result appears.
In IB or more demanding IGCSE work, the chain may be longer still.
If every intermediate line is crowded together, notation changes casually and equal signs are used to mean several different things, then even a knowledgeable student can struggle to diagnose her own work.
Neatness, in this sense, is not decoration.
It is fault tolerance.
A readable page allows the student to return to the last place where the Mathematics was known to be sound.
The student who asks, “Where did I go wrong?”
There is nothing wrong with asking for help.
But I listen carefully to the form of the question.
“Where did I go wrong?” can mean two very different things.
Sometimes it means: “Please point to the line so I can correct it.”
Sometimes it means: “Help me learn how to find the line myself.”
The second is more valuable.
If I simply tap the erroneous line every time, the correction is quick. We save time. The book looks tidy.
But I have also taken over the diagnostic work.
A better intervention may be: “Which line are you still confident is true?”
Or: “Can you substitute your answer back into the original equation?”
Or: “What should the sign of this quantity roughly be?”
Or: “Did the mistake happen before or after differentiation?”
These questions narrow the search without performing it completely.
The aim is not to make error correction unnecessarily difficult.
The aim is to teach the student how to investigate her own Mathematics.
An error is easier to repair when the answer has expectations
Students become better at detecting mistakes when they have some expectation of what a sensible answer should look like.
This is one of the least glamorous but most useful habits in Mathematics.
Before calculating, ask:
- Should the answer be positive or negative?
- Should it be larger or smaller than the starting quantity?
- Should the probability be close to zero, close to one, or somewhere in between?
- If this is a gradient, does the graph appear to rise or fall?
- If this is a length, is the scale plausible?
- If this is an angle in the diagram, would 147° make visual sense?
These are not substitutes for calculation.
They are constraints around calculation.
Imagine a student solving a percentage-growth problem in which an amount rises from $800 by 15%.
If the calculator returns $92,000, the calculator itself has not necessarily failed.
The student’s relationship with the result has failed.
A mathematically mature student does not accept a number simply because a machine displayed it.
The result has to re-enter the world of the question.
I like students who argue with their calculators
Not performatively.
Not because calculators are unreliable.
Because the student has enough mathematical expectation to notice when the output deserves investigation.
A calculator is excellent at carrying out the instruction it receives.
It is much less useful at telling a student whether the instruction should have been given.
This distinction becomes important in Secondary Mathematics.
When the student’s conceptual picture is weak, the calculator can create a strange illusion of certainty.
The display says: −3.7421.
The number has four decimal places.
It looks official.
But if the question asks for the side length of a triangle, the negative sign should create immediate discomfort.
That discomfort is useful.
It means the student has not surrendered meaning to computation.
Not every wrong answer deserves the same amount of attention
There is also a danger in turning every small error into a philosophical investigation.
Sometimes a mistake is genuinely trivial.
A student copies 17 as 71, notices it, corrects it and continues.
We do not need a twenty-minute lesson on transposition.
Good teaching requires proportion.
What I am looking for is recurrence.
Does the same type of failure appear repeatedly?
Does it survive correction?
Does it appear only under time pressure?
Does it disappear in topical work but return in mixed papers?
Does the student understand the explanation today but reproduce the same error next week?
A single error is an event.
A recurring error is a pattern.
Patterns deserve attention.
The correction book should not become a museum
Parents sometimes show me beautifully maintained correction books.
Every wrong question has been recopied.
Every correct solution is written neatly underneath.
The pages are immaculate.
But there is one question I want to know:
Did the correction change what the student does next time?
A correction that merely replaces a wrong solution with a correct one can become a kind of archival activity.
The book records that an error once existed.
It does not necessarily reduce the probability of its return.
For correction to become useful, the student has to extract something.
Perhaps: “When differentiating a product, I keep treating it as two separate derivatives.”
Or: “I recognise the trigonometric identity, but I choose it only when the question looks exactly like the example.”
Or: “My method is sound, but I tend to lose the negative when substituting a negative value into a squared term.”
Then we can construct another question that tests the same vulnerability in a different form.
That second encounter matters.
Repair is not demonstrated by rewriting the old question correctly while the model answer is visible.
Repair is demonstrated when the student meets the same underlying difficulty again and behaves differently.
This is where transfer enters correction
Suppose a student makes the error (a + b)² = a² + b².
The teacher explains the expansion.
The student corrects the question.
Fine.
But we do not yet know whether the idea is repaired.
A week later, the student sees (2x − 3)² inside a different problem.
Does the incorrect shortcut return?
If not, the correction has begun to transfer.
This is why the useful life of an error extends beyond the page on which it occurred.
A good correction should change future behaviour.
Otherwise we have only improved the historical record.
What I notice when students erase
There is one small physical habit I have become interested in over the years.
Erasing.
Some students erase the instant they hear that something is wrong.
The line disappears before we have looked at it.
It is almost emotional.
Wrong work feels untidy, and therefore must be removed.
I often ask them to leave it for a moment.
Not because I want pages full of mistakes.
Because the wrong line belongs to the lesson too.
There is a useful difference between: “I was wrong.” and: “This particular relationship was wrong, and now I understand why.”
The first can become a judgement about the student.
The second remains a judgement about the Mathematics.
That separation is healthy.
Adolescents can take mathematical error personally
By Secondary school, students have often developed a fairly clear story about themselves.
“I am a Math person.”
“I am not a Math person.”
“I was always good at Math.”
“I have never been good at Math.”
These identities are stronger than they appear.
A student who has always been regarded as strong may find error unexpectedly threatening.
The problem is no longer merely that Question 6 is wrong.
The wrong answer conflicts with a personal identity: “I am supposed to be good at this.”
That can produce odd behaviour.
Rushing past corrections.
Blaming the question.
Calling everything careless.
Avoiding difficult work.
Preferring familiar exercises that preserve the feeling of competence.
For these students, learning to investigate an error calmly is more than an examination technique.
It is a way of separating competence from perfection.
Strong mathematicians are not people who never encounter wrong paths.
They are people who have learned how to detect, inspect and abandon them.
The weaker student has a different problem
A student who has struggled for a long time may respond to errors differently.
Instead of protecting an identity of competence, she may already expect failure.
When an answer is wrong, the reaction is immediate: “See? I can’t do Math.”
This is equally unhelpful.
A wrong answer contains far less information than that sentence claims.
If the student selected the correct method, carried out six correct steps and lost one mark through a sign error, “I can’t do Math” is an extraordinarily inaccurate summary of the evidence.
The useful response is more precise: “You understood almost the entire question. The failure occurred here.”
That is not empty encouragement.
It is diagnosis.
And accurate diagnosis is often more reassuring than generic reassurance because it turns an apparently enormous problem into something bounded.
Parents can help by changing one question
When a child brings home a disappointing Mathematics paper, there are many reasonable things a parent might ask.
“How many marks did you get?”
“Did you study enough?”
“Was the paper difficult?”
“How did your friends do?”
I would add another:
“What kinds of mistakes were they?”
Not accusatorily.
Curiously.
Were the lost marks mostly because concepts were not understood?
Because the child could not recognise which method to use?
Because working became inaccurate?
Because time ran out?
Because the final instruction was missed?
A 58 can be produced in many ways.
A 78 can be produced in many ways too.
The number matters, but the anatomy of the number tells us what to do next.
Sometimes the solution is not more tuition
This matters particularly when families are deciding whether a child needs additional support.
A poor examination result can create urgency.
The understandable response is to add lessons, add papers, add homework.
Sometimes that is exactly what is needed.
But if the student already understands the concepts and the dominant problem is careless execution under timed conditions, another three hours of explanation may not be the best first repair.
If the student executes flawlessly once a method is provided but cannot recognise methods independently, doing more chapter-by-chapter worksheets may produce a false sense of improvement.
If the student is accurate but very slow, the problem may be fluency and decision cost.
If the student is anxious and repeatedly second-guesses correct work, sheer volume may worsen the problem.
Before increasing the amount of tuition, it is worth identifying what additional tuition is supposed to change.
More is useful when it is more of the right thing.
This is one reason I prefer small-group teaching
There is a part of error behaviour that is difficult to see in a large room.
You have to watch the student.
Not merely the page.
How quickly does she notice something is wrong?
Does she check the previous line or start again?
Does she ask for the answer or ask for a clue?
Does she accept an implausible result?
Does she become flustered after the first error and create three more?
Does she recover?
These observations are ordinary. They are not dramatic teaching moments.
But over time they form a picture of how the student thinks when Mathematics stops going smoothly.
And that picture helps determine what kind of support is actually useful.
The examination is partly a recovery test
We often describe examinations as tests of knowledge.
They are.
But Mathematics examinations also test recovery.
A student misreads a line.
A method fails.
A calculator result looks strange.
A difficult question consumes too much time.
The paper does not stop.
What happens next matters.
Can the student abandon a route without becoming emotionally stuck to it?
Can she leave a difficult question, preserve the marks elsewhere, and return later?
Can she detect that an answer is implausible?
Can she restart from the last trustworthy line instead of from the beginning?
Can she remain mathematically organised after one error?
These are not separate from mathematical competence.
They are part of making competence usable under examination conditions.
Good correction becomes faster with maturity
Early on, a student may need a teacher to help locate the error.
Later, she should begin narrowing it herself.
The progression might look something like this.
First: “I got it wrong.”
Then: “My algebra went wrong.”
Later: “The method is correct, but something happened between these two lines.”
Eventually: “I substituted a negative value incorrectly when squaring; the rest of the method is fine.”
That increasing resolution is important.
The student is learning to inspect her own thinking with greater accuracy.
And once the fault can be located accurately, repair becomes much cheaper.
The aim is not a student who never makes mistakes
I do not think that is a sensible educational objective.
The harder the Mathematics becomes, the more likely a student is to encounter moments where an assumption fails, a manipulation goes wrong, or a first method turns out not to be useful.
A student who never makes mistakes may simply be doing work that is too familiar.
What I want instead is a student whose mistakes become increasingly informative.
A student who can look at a wrong answer without immediately turning it into a judgement of intelligence.
A student who preserves enough working to reconstruct what happened.
A student who can tell the difference between not understanding the idea and mishandling one line.
A student who corrects not merely the page, but the habit that produced the page.
And finally, a student who can meet the same underlying difficulty in a new form and not repeat it.
That is a much more durable kind of progress.
What good tuition should leave behind
There is a simple test I like.
Imagine the tutor is not there.
The student is sitting alone in an examination hall.
Halfway through a question, the answer no longer makes sense.
What happens?
If good teaching has taken root, the student has something better than memory of a model solution.
She has a way to proceed.
- Return to what is known.
- Find the last trustworthy line.
- Check the mathematical relationship.
- Test the answer against the question.
- Change the method if necessary.
- Preserve composure.
- Continue.
That is the kind of independence tuition should be building.
Not the ability to produce perfect pages while somebody knowledgeable is standing beside the desk.
The ability to recover when nobody is.
The quiet lesson inside a wrong answer
We understandably celebrate correct answers.
They are evidence of learning, and examinations reward them.
But I have come to think that the way a student handles a wrong answer can tell us something a correct answer sometimes cannot.
A correct answer shows that the route worked.
An error followed by intelligent diagnosis shows that the student can inspect the route itself.
That ability becomes increasingly valuable as Mathematics grows less predictable.
So when a student looks up and says, “I think this is wrong,” I do not regard the question as finished.
Quite often, that is when the most interesting part begins.
Because Mathematics is not only learning how to be right.
It is learning how to find your way back when you are not.