Most Mathematics learning asks the student to answer questions.

The textbook decides the problem. The teacher decides the numbers. The examination decides what information is given, what must be found and which conditions matter.

The learner responds.

There is another way to test mathematical understanding.

Ask the student to design the question.

Creating a valid Mathematics problem requires more than remembering a procedure. The learner has to understand what makes the method possible, what information is necessary, what conditions must be preserved, what changes would make the question easier or harder and what an answer should look like.

Problem posing turns the student from a passenger into an engineer of the mathematical structure.

The Short Answer

To test whether you understand a topic, create a question that uses the idea, solve it, and then change one feature at a time.

Change the numbers without changing the method.
Change the wording while preserving the relationship.
Change one condition so a different method becomes necessary.
Create a near-miss where the familiar method no longer applies.
Work backwards from a desired answer.
Create an easier and a harder version of the same structure.

A useful cycle is:

Solve → Identify structure → Design → Test → Vary → Break → Repair → Explain.

Why Creating a Question Is Harder Than Answering One

When you answer a question, much of the mathematical architecture has already been built for you.

The quantities exist. The conditions are supplied. The target has been chosen. Often the chapter heading hints at the method.

When you create a question, you must decide all of these things yourself.

You have to know what information is sufficient. You have to avoid contradictions. You have to make sure the question actually has an answer. You have to decide whether the answer is unique. You have to ensure the method you intend is genuinely valid.

That is why question creation can expose gaps that ordinary practice hides.

Begin by Reverse-Engineering a Solved Problem

Take one question you already know how to solve.

Do not copy the solution.

Ask what made the solution possible.

What mathematical object was present?
What information was essential?
What condition triggered the method?
What information could be changed without changing the route?
What change would destroy the route?

Now write a new question with the same hidden structure.

This is much more useful than changing 5 to 7 and calling it a new problem. The goal is to preserve the mathematical architecture while changing the surface.

Create a Same-Method Variation

Suppose the original problem requires simultaneous equations because two unknown quantities satisfy two independent linear conditions.

Create a new story.

Instead of adult and child tickets, use two types of stationery. Instead of total tickets and total revenue, use total items and total cost.

The numbers can change. The context can change. The hidden structure remains two unknowns plus two independent linear constraints.

If you can design several such problems, you are beginning to understand the family rather than one page.

Create a Different-Method Variation

Now make one deliberate change that forces another route.

Perhaps one of the two conditions becomes nonlinear.

Perhaps only one equation remains, making the problem underdetermined.

Perhaps the target changes from finding the unknown values to determining whether a unique solution exists.

This exercise teaches a deeper question:

What exactly makes my original method appropriate?

Work Backwards from an Answer

Another powerful design strategy starts from the destination.

Choose an answer first.

Suppose you want a quadratic equation with roots 3 and 5.

Build the factors (x − 3)(x − 5), expand them, and then create a context or algebraic question that leads to that equation.

Now reverse-design another quadratic with a repeated root. Then another with no real roots.

Working backwards turns formulas and conditions into construction tools.

Create an Example and a Near-Miss

If you are learning a definition, create one object that satisfies it and another that almost satisfies it.

For direct proportion, create a straight-line graph through the origin and another straight-line graph with a non-zero intercept.

For a function, create a valid mapping and another relation where one input has two outputs.

For a right-angled triangle method, create one triangle with a 90-degree angle and another with angles of 89, 60 and 31 degrees.

The near-miss teaches the boundary better than another perfect example.

Create a Counterexample Question

Write a false general claim that sounds plausible.

Then ask the student to find a counterexample.

“If two quantities both increase, they must be directly proportional.”
“If a quadratic has a positive leading coefficient, it must have two real roots.”
“If two triangles have two equal angles, they must be congruent.”

Designing such questions requires understanding both the rule and the temptation that causes people to overgeneralise it.

Create an Easier Version

After solving a difficult question, design a version for a learner one stage earlier.

What complexity can be removed while preserving the central idea?

Use friendlier numbers. Reduce the number of steps. Make the representation more explicit. Remove one distractor.

This tests whether you know which part of the original problem carries the real mathematical load.

Create a Harder Version

Now add difficulty deliberately.

Hide the topic label.
Add irrelevant information.
Change friendly numbers to awkward values.
Require an intermediate quantity before the final target.
Combine the topic with another topic.
Ask for explanation rather than only calculation.

Difficulty should come from a meaningful new demand, not random ugliness.

Create a Question That Has Insufficient Information

Students are conditioned to believe every school question must have enough data.

Create one that does not.

Then ask what information is missing.

This teaches data sufficiency.

The learner must understand not only how to solve the problem but what would be necessary for a unique solution to exist.

Create a Question With Redundant Information

Now do the opposite.

Include one true fact that is not needed for the main route.

Ask another learner to solve without announcing that a distractor exists.

This trains the distinction between information that is present and information that is relevant.

Create a Multi-Part Question

Multi-part questions reveal mathematical dependency.

Design part (a) so it establishes a result that part (b) can use. Let part (c) generalise or interpret the earlier work.

Now ask:

Could part (b) be solved independently?
Is the handoff from one part to the next clear?
Does each part increase the mathematical demand?

Question design becomes a study of dependency architecture.

Create a Question That Forces Representation

Some questions can be solved by immediate calculation. Others should require translation first.

Design a word problem where the learner must define a variable, draw a diagram, form a table or build an equation before the method becomes visible.

If the answer can be guessed directly from the numbers, the question may not test the intended representation skill.

Create a Question That Tests Recognition

Do not name the method in the title.

Instead, place enough structural evidence inside the question for the learner to recognise the route.

This is very different from writing “Solve using simultaneous equations”.

Recognition itself becomes part of the assessment.

Create a Question That Tests Checking

Design a problem where an answer can be independently verified.

Perhaps substitution checks the equation.

Perhaps a graph confirms an algebraic solution.

Perhaps units or magnitude can reject an impossible answer.

Then require the solver to state the check, not merely the answer.

Mark Your Own Question

A well-designed problem should survive its own marking scheme.

Solve it completely.

Check that the answer exists.

Check whether multiple valid answers are possible.

Check whether the intended method is necessary or merely one possible route.

Check whether any hidden assumption makes the question ambiguous.

This process often teaches more than solving another textbook item.

Use Parameters to Generate Question Families

Once one question is valid, replace one fixed value with a parameter.

For which parameter values does the question have one solution, two solutions or no solution?

For which values does the method change?

Question generation now becomes family generation.

Additional Mathematics Benefits Strongly from Problem Posing

A-Math contains many structures that become clearer when learners construct their own examples.

Create quadratics with chosen root behaviour.
Create functions that are one-to-one on a chosen domain.
Create trigonometric identities by starting from a known identity and transforming it.
Create optimisation problems whose stationary point has a chosen interpretation.
Create logarithmic equations with controlled domains and valid solutions.

Construction forces conditions to become explicit.

A Problem-Posing Routine

Solve one reliable example.
Name the hidden mathematical structure.
List the conditions that make the method valid.
Create a new same-structure question.
Create one near-miss where a condition fails.
Create an easier version.
Create a harder version.
Solve and mark every question you created.
Explain what changed and why.

What Parents Can Notice

A learner who can create valid Mathematics questions is beginning to demonstrate ownership.

The child may say, “If I change this condition, the method no longer works,” or “I can make a harder version by hiding the intermediate quantity.”

Those statements reveal structural understanding that a worksheet score alone may not show.

What Tutors Should Do

Occasionally stop being the only person who writes the questions.

Ask the student to design one.

Then inspect the design together.

Is there enough information? Is the target clear? Does the intended method really follow? What change would create a different route?

The discussion turns assessment into mathematical reasoning.

Final Answer

How can creating your own Mathematics questions improve learning?

Because question design forces you to understand the conditions beneath the procedure. You must decide what information is necessary, what target is meaningful, what method is valid, what changes preserve the structure and what changes break it.

Start with a solved problem. Reverse-engineer its structure. Create a same-method variation, a near-miss, an easier version and a harder version. Solve everything you create.

When you can design a problem deliberately, you are no longer only recognising Mathematics made by somebody else. You are beginning to control the mathematics itself.

Continue the How to Learn Mathematics Series

Continue learning mathematics

Mathematics Learning Library.

Small Group Tutorials

How to Learn Mathematics | Create Your Own Questions to Test Understanding