There is a question I sometimes ask students that sounds slightly unfair.
We have not started the calculation.
The calculator is still on the table.
The algebra has not been written.
And I ask: “What do you think the answer should be roughly?”
Students often look at me as though I have reversed the order of Mathematics.
“How would I know? I haven’t worked it out yet.”
That response is understandable.
School Mathematics trains students, quite properly, to calculate accurately. We substitute values, manipulate equations, apply formulas and obtain exact answers.
But over the years, I have become increasingly interested in what students know before the calculation begins.
Should the answer be large or small?
Positive or negative?
Closer to 2 or closer to 200?
Should the graph be rising or falling?
Should the probability be nearer zero or nearer one?
Should this length be shorter than the side already shown in the diagram?
These are not substitutes for Mathematics.
They are part of Mathematics.
And when students never develop this layer of judgement, calculation can become strangely detached from meaning.
The short answer
I ask students to estimate before calculating because a number should not arrive on the page without expectations.
The student should have some idea of the territory in which a sensible answer can live.
That expectation acts as a boundary around the calculation.
It helps detect errors.
It improves method choice.
It reduces blind reliance on calculators.
And, perhaps most importantly, it encourages the student to remain connected to the problem while doing the mathematics.
A student who knows only how to obtain an answer is less secure than a student who also knows what sort of answer ought to be possible.
The calculator can be perfectly correct and the student completely wrong
This is an important distinction.
Students sometimes say: “But the calculator gave me that.”
The calculator almost certainly did.
Calculators are very reliable at answering the question they are given.
The problem is that the student may have given it the wrong question.
A missing bracket.
A degree-radian setting.
An incorrect negative sign.
A percentage entered as 15 instead of 0.15.
A value copied wrongly.
A formula arranged incorrectly before anything was keyed in.
The calculator carries out the instruction faithfully.
It cannot protect the student from the instruction itself.
That protection has to come from somewhere else.
One source is mathematical expectation.
If a student expects an answer around 800 and the display says 79,342, something should feel wrong before the answer reaches the page.
That feeling is valuable.
Mathematics should create discomfort sometimes
I like the moment when a student looks at a result and hesitates.
“This seems too big.”
Good.
Now we have Mathematics happening beyond procedure.
The student is comparing the output against a mental model of the question.
Suppose a $600 item increases in price by 10%.
A student mistakenly enters something and obtains $6,600.
There is no need for advanced mathematics to detect the problem.
Ten per cent is a relatively small increase.
A sensible answer should be somewhat larger than $600.
Perhaps around $660.
The exact calculation still matters.
But estimation has already constructed a fence around the answer.
$6,600 sits far outside it.
Without that fence, the student may simply copy what appears on the screen.
Estimation is not guessing
Students occasionally resist estimating because they think Mathematics should be exact.
They are right about the value of exactness.
But estimation is not the opposite of exact mathematics.
It is a different mathematical task.
Consider: 19.8 × 5.1.
Before calculating, I might think: 20 × 5 is about 100.
Therefore the exact answer should be somewhere around 100.
I am not claiming that 100 is the answer.
I am establishing scale.
If the calculator later displays 100.98, that feels plausible.
If it displays 1,009.8, something deserves inspection.
The estimate has done its job.
It has not replaced the calculation.
It has created a standard against which the calculation can be judged.
Strong students do this more often than they realise
One reason estimation can be difficult to teach is that experienced mathematical thinkers often perform it almost invisibly.
They see an equation and already expect the sign of the answer.
They look at a graph and know roughly where a root should lie.
They read a geometry question and understand that one suggested angle is impossible.
They see an exponential model and know that a linear-looking answer deserves suspicion.
This expectation becomes so natural that the expert may forget it is a learned capability.
Students often see only the formal solution that comes afterwards.
They see the teacher differentiate.
Substitute.
Simplify.
Write the answer.
They do not always see the silent prediction that preceded the first line.
A-Math makes this particularly valuable
Additional Mathematics gives students more powerful machinery.
And powerful machinery makes judgement more important, not less.
Take differentiation.
Suppose a curve is visibly falling at a particular point.
The derivative there should be negative.
That expectation exists before any substitution.
If the student calculates a positive gradient, there are now two possibilities.
The graph has been misunderstood.
Or the calculation is wrong.
Either way, the disagreement tells us where to investigate.
Without the expectation, a positive number can pass through unnoticed.
The student may have executed every button press accurately and still failed to connect the answer back to the curve.
Geometry has expectations too
Imagine a triangle with two sides of 6 cm and 8 cm.
A student calculates the third side as 27 cm.
Even before checking the working, something should trouble us.
The triangle inequality tells us that the third side must be less than 14 cm.
This is not estimation in the casual sense.
It is structural knowledge constraining the answer.
Mathematics often contains these boundaries.
Probabilities lie between 0 and 1.
Lengths are non-negative.
Squared quantities cannot be negative in ordinary real-number contexts.
Angles obey geometric constraints.
Coordinates must fit the graph.
Solutions may have domain restrictions.
The stronger the student’s conceptual understanding, the richer these expectations become.
This is where conceptual understanding protects execution
A student can make an arithmetic error and still be protected by conceptual knowledge.
That is one of the reasons I do not like treating concepts and accuracy as completely separate worlds.
Understanding can catch calculation.
Suppose a student is solving an equation where the original expression is clearly positive over the relevant interval.
If the final answer implies something impossible under that condition, conceptual understanding should object.
The student does not need to know immediately where the error occurred.
It is enough initially to know that the answer cannot be accepted yet.
That refusal is an important mathematical act.
Some students trust precision more than reason
A calculator display can be psychologically persuasive.
3.71842861 looks serious.
It has many digits.
It feels more authoritative than: “Probably around 4.”
Yet the long number may come from a completely incorrect setup.
Precision and correctness are not the same thing.
This is a lesson worth learning early.
A highly precise wrong answer remains wrong.
A rough but structurally sensible estimate can sometimes contain more mathematical understanding.
Students need both.
The estimate sets the world.
The calculation refines it.
The danger begins when students stop asking what numbers mean
I sometimes see students complete a calculation and immediately move to the next question.
No pause.
No interpretation.
The number is treated as the endpoint.
But most examination questions are not really asking for floating numbers.
They are asking for something in a world.
A distance.
A time.
A gradient.
A probability.
A percentage.
A coordinate.
An area.
A rate of change.
A number only becomes an answer when it re-enters that world.
If the question asks for the number of people, 32.7 is not a final answer.
If the question asks for a probability, 1.4 cannot simply be copied down.
If the question asks for a physical length, a negative value requires interpretation or rejection.
Meaning still has authority over calculation.
One useful habit: predict the sign
This is a very small practice.
Before calculating, ask: positive or negative?
It works surprisingly often.
In coordinate geometry, which direction should the gradient take?
In differentiation, is the function increasing or decreasing?
In financial change, is the amount gaining or losing?
In vectors, which direction should the component point?
In algebra, should a rearranged quantity logically remain positive under the given conditions?
The prediction takes seconds.
But it gives the student an immediate diagnostic check once the calculation is complete.
Another useful habit: predict the scale
Sometimes the sign is obvious, but the size matters.
Take: 298 × 21.
A student need not know the exact answer immediately.
But 300 × 20 is 6,000.
So the final result should be somewhere around 6,000.
Not 600.
Not 60,000.
That simple estimate protects place value.
It also helps students understand whether the exact answer is reasonable.
The same habit becomes more sophisticated later.
If a quadratic graph crosses the x-axis around x = 3 from the diagram, a calculated root of 300 deserves scrutiny.
If a percentage change is small, the final quantity should not suddenly be ten times larger.
Scale is part of mathematical meaning.
Why this matters more with AI
Students now have extraordinary access to worked answers.
A photograph of a question can produce a solution in seconds.
That is useful.
It also changes what students need to be good at.
When answers become easier to obtain, the ability to judge answers becomes more valuable.
Does this method match the syllabus?
Has the problem been interpreted correctly?
Does the answer satisfy the original conditions?
Is the result plausible?
Has the system solved a slightly different question?
A student who has no expectation of the answer has very little basis for evaluating an external solution.
She can compare only authority.
“My answer says this.”
“The AI says that.”
“The answer key says something else.”
Which one wins?
Mathematical judgement gives the student another source of evidence.
Estimation can expose shallow formula use
Suppose a student calculates the area of a circle with radius 5 cm and obtains approximately 15.7 cm².
The formula may have been remembered.
The calculator may have been used.
But something is off.
A square of side 5 already has area 25 cm².
A circle of radius 5 is much larger than that square.
So 15.7 cm² should feel suspicious even before we inspect the formula entry.
This is useful because it tells us the difficulty is not merely calculator accuracy.
The student may not have a strong enough spatial model of what the calculation represents.
Estimation exposes that.
Not every question needs an estimate
There is a boundary here.
We should not turn estimation into another ritual.
Students already have enough rituals.
“Estimate before every question” would be mechanical and wasteful.
Sometimes the answer is symbolic.
Sometimes the problem is too abstract for a useful numerical expectation.
Sometimes the calculation is trivial.
Sometimes estimating takes longer than solving.
The useful principle is not that every calculation needs a written approximation.
It is that students should develop the habit of asking whether the problem provides any useful constraints before accepting the result.
Sometimes the constraint is size.
Sometimes sign.
Sometimes shape.
Sometimes domain.
Sometimes simply common sense.
There is also bad estimation
A student can estimate carelessly.
“Looks about right.”
That is not enough.
Useful estimation should have a reason.
Why around 100?
Because 19.8 is close to 20 and 5.1 is close to 5.
Why should the gradient be negative?
Because the graph is falling as x increases.
Why should the probability be fairly small?
Because the desired outcomes form a small part of the sample space.
The estimate does not need elaborate working.
But it should be anchored.
Otherwise we replace blind calculation with blind intuition.
That is no improvement.
What I do when the exact answer and expectation disagree
I do not immediately tell the student which one is wrong.
The disagreement is useful.
We inspect both.
Was the expectation based on a valid idea?
Was the calculation entered correctly?
Was the formula appropriate?
Was the diagram interpreted properly?
Sometimes the estimate was poor.
That matters too.
Mathematical judgement should itself be correctable.
The purpose is not to teach students to trust intuition over calculation.
The purpose is to make the two systems talk to each other.
When they agree, confidence increases.
When they disagree, investigation begins.
This becomes a form of self-checking
Students are often told: “Check your work.”
But that instruction can be too vague.
Check what?
Redo every question?
Press the calculator buttons again?
Read the answer twice?
Estimation gives checking a specific starting point.
Does the answer have the expected sign?
Is the magnitude plausible?
Does it fit the graph?
Does it satisfy the equation?
Does it respect the conditions?
That is much more useful than scanning a page hoping to notice something.
Parents can use this without teaching the syllabus
This is one of the easiest mathematical habits for parents to encourage without having to remember Secondary Mathematics.
If your child shows you an answer, you can ask: “Does that number seem reasonable?”
Then: “Why?”
You do not need to know the correct answer.
You are asking the student to connect calculation back to meaning.
If a teenager says a taxi travelling for ten minutes covered 430 kilometres, you do not need algebra to know that something deserves attention.
If a percentage discount makes an item more expensive, something has broken.
If a probability is 140%, something needs interpretation.
Parents can encourage mathematical judgement without becoming the Mathematics teacher.
The student who never estimates often works harder than necessary
This may sound surprising.
Estimation can save time.
A student who blindly follows a wrong route may spend five minutes completing it.
A student with an expectation may notice within thirty seconds that the route is producing impossible values.
Similarly, a student can use scale to choose methods.
If an exact answer is not required, an approximate comparison may be enough.
If answer options are widely separated, estimation can eliminate impossible choices quickly.
Judgement can reduce computation.
Estimation is also useful for examination recovery
Under timed conditions, students will make mistakes.
The question is how quickly the mistake becomes visible.
A student with strong expectations has more alarms.
The answer is too large.
The sign is wrong.
The graph and gradient disagree.
The probability is impossible.
The coordinate lies nowhere near the diagram.
These alarms help the student catch an error before moving on.
That can save marks.
More importantly, it can prevent one wrong intermediate value from contaminating the rest of a multi-part question.
How I know this capability is developing
I listen to the student’s language.
Early on: “The calculator says 47.2.”
Later: “I got 47.2, which seems reasonable because I expected something around 50.”
Or: “I got −3, but the gradient should be positive here, so I think I made an error.”
Or: “This root works algebraically, but it is outside the domain.”
Or: “The value is technically possible, but it does not fit the scale of the graph.”
Those sentences tell me the student is no longer treating the final number as an unquestionable object.
She is judging it.
This is a different kind of confidence
Confidence in Mathematics should not mean: “I am sure my answer is right.”
That is often impossible.
A more useful confidence is: “I know how to decide whether my answer deserves trust.”
That confidence has evidence beneath it.
The student knows what to check.
She knows what should roughly be true.
She can identify disagreement.
She can return to the working.
This is much more robust than simply feeling certain.
Good tuition should eventually create an internal examiner
Not an anxious one.
Not a voice that constantly says, “You are probably wrong.”
Something calmer.
A small internal question: “Does this make sense?”
That question should survive after the tutor disappears.
It should survive after the notes close.
It should survive after the worked examples are gone.
It should be present in the examination hall.
And eventually, it should become part of how the student approaches quantitative information beyond school.
Mathematics outside school depends on this too
Adults rarely solve textbook exercises in daily life.
But we estimate constantly.
Is this bill plausible?
Is this interest rate meaningful?
Can this journey really take twelve minutes?
Does this graph’s scale exaggerate the change?
Is this percentage being interpreted correctly?
Is a claimed statistic even possible?
The ability to calculate is useful.
The ability to notice that a number does not belong in the world being described is equally valuable.
Perhaps more so when someone else performed the calculation.
What parents might watch for
When your child finishes a Mathematics question, notice whether the answer ends the thinking immediately.
Does she ask whether it makes sense?
Can she tell you whether it should be large or small?
Can she explain the sign?
Does she notice impossible probabilities or physical quantities?
Does she challenge the calculator occasionally?
Can she identify a rough range before obtaining the precise value?
These are quiet signs.
They do not necessarily raise marks overnight.
But they show that the Mathematics is becoming connected.
The deeper reason I ask
I do not really care whether a student can guess the answer.
Guessing is not the point.
I care whether the student enters a calculation with a model of the situation still alive.
What is changing?
In which direction?
By roughly how much?
What is possible?
What is impossible?
What would surprise me?
These questions keep meaning attached to method.
Without them, Mathematics can become a tunnel.
The student enters through a formula, performs a sequence of steps and emerges carrying a number.
The number may be right.
But the student may never have looked outside the tunnel.
The quiet habit
So sometimes, before a student reaches for the calculator, I ask: “Roughly?”
At first, the question can feel irritating.
The exact answer seems only a few buttons away.
But eventually, some students begin doing it without being asked.
They pause.
Look at the question.
Set a rough expectation.
Then calculate.
And when the answer appears, they do not simply accept it.
They compare.
That small comparison is one of the places where Mathematics stops being only a procedure and becomes judgement.
Because a student should not merely be able to produce numbers.
She should gradually learn which numbers deserve to be believed.