Did you know? Two students can earn exactly the same mark on a mathematics paper yet need completely different help. One student understands algebra but makes a careless sign error. Another has memorised the shape of several examples without understanding equality. The score reports the same number. Their next lesson should not be the same.
This is one way examinations fail—not because exams are automatically bad, but because a result is mistaken for a complete explanation of what someone knows, how they think and what they can do next.
One-minute answer: an examination can mislead when questions sample too narrow a range of skills, grading misses important reasoning, time pressure or access barriers distort performance, or a single score replaces useful feedback. A valid test of selected outcomes is not a measure of every human capability. The remedy is to design assessments for a clear purpose and use additional evidence when important decisions depend on the result.
A test can be useful and incomplete at the same time
An exam is a structured way to gather evidence about performance on selected tasks under selected conditions. It may help with certification, placement, identifying attainment or checking whether taught material can be used. The limitation is that no finite paper can observe everything a learner might know or become capable of doing.
Some skills are suited to written responses; others require an oral explanation, a laboratory procedure, a project or observation across time. A student’s ability to discuss a source thoughtfully cannot always be reduced to the number of facts recalled in a short-answer question.
Singapore’s Ministry of Education says in the Primary Mathematics syllabus that assessments should provide timely, rich feedback to students and teachers. It distinguishes assessment of learning from assessment for learning. This is a useful foundation: collecting a mark and improving a learner’s understanding are related but different activities.
The six breaks that can occur in an examination system
- The purpose becomes unclear. A paper meant to diagnose misconceptions is used as though it were a final verdict on talent, motivation or future opportunity.
- The sample becomes too narrow. Too few tasks, question types or contexts are used to support a much broader claim about ability.
- Technique replaces understanding. A student recognises repeated templates but cannot select an appropriate method when wording or circumstances change.
- The conditions distort performance. Access barriers, unclear instructions or inappropriate conditions affect what a student can show.
- Marks conceal the cause of mistakes. A score does not distinguish an arithmetic slip from an insecure concept, a reading problem or a missing step.
- Feedback never changes teaching. Results are filed, ranked or celebrated without any response to the needs they revealed.
These are different failure mechanisms. A test can have carefully designed marking rules yet still be used for the wrong decision. Conversely, a low score may be a sound signal that important teaching or practice is needed. Investigate the specific purpose and evidence before declaring an examination unfair or meaningless.
The trap of teaching only the pattern in the past paper
Illustrative example. A Secondary 1 learner practises ten equations of the form 3x + 2 = 14. They learn to subtract two and then divide by three. They get every practice question correct.
The next test asks: ‘A box holds three identical bundles and two spare pencils. There are fourteen pencils altogether. How many pencils are in each bundle?’ The student freezes because the equation is not already written.
The original practice assessed competence at applying a known routine. The new problem also requires forming the mathematical relationship. Those are related, but not identical, skills.
A useful follow-up is not simply another set of near-identical equations. Ask the student to explain what x means, write the equation themselves, solve it and check the result in the story. Then see whether they can repeat that reasoning with a changed context.
The unchanged Secondary 1 Mathematics Tutor Clementi guide explains the transition from arithmetic to algebra. It provides a practical example of why a tutor needs to see the reasoning behind a student’s answer rather than just the mark.
High marks do not prove every skill; low marks do not explain every cause
Consider a student who earns 82%. That number tells us about performance according to a particular mark scheme, on that paper, under those conditions. It does not independently demonstrate strong transfer to unfamiliar situations, creativity, motivation or judgement.
Likewise, a score of 48% does not tell us whether errors came from misunderstandings, missing prior knowledge, time management, reading difficulty, ineffective revision or other circumstances. The paper itself can provide clues if someone reviews the actual working.
UNESCO’s overview of learning assessments discusses different purposes for assessments, including informing teaching and evaluating learning at the end of a period. The appropriate interpretation of a result depends on which of those purposes the assessment was designed to serve.
When the mark scheme rewards the wrong evidence
If a teacher wants to know whether a student can reason with evidence, but only rewards memorised phrases, a high mark may reflect good recall without demonstrating that intended reasoning. The assessment and the learning objective are misaligned.
But there is no one perfect exam format. Well-designed short answers are useful for some knowledge and skills, and longer essays or projects can introduce their own reliability, fairness and marking challenges.
The solution is to choose tasks suited to the learning goal, make success criteria clear and avoid drawing claims beyond what the task actually measures. Where practical, use multiple examples or assessment methods to build a more reliable picture.
Why feedback matters more than another percentage
The Education Endowment Foundation’s guidance on teacher feedback stresses high-quality instruction and formative assessment before feedback, as well as information that learners can use to close a gap. Merely delivering more written comments or marks is not the same as effective feedback.
Suppose the teacher writes, ‘Incorrect—revise fractions.’ That tells the child where the broad problem lies but little about the first wrong step. A better response might say, ‘You added the numerators and denominators separately. Explain why the parts must describe the same whole before adding.’
Now the student has a concrete action. A follow-up question can show whether the correction worked. The feedback has become a short learning loop rather than a mark with extra words attached.
An examination can be stressful without necessarily being invalid
Timed assessments deliberately test some aspects of performance under time constraints. For some qualifications, that is relevant. For others, speed is a poor substitute for the intended knowledge or professional skill.
A fair assessment asks whether the conditions fit its purpose and whether eligible learners receive appropriate access arrangements. The mere existence of pressure does not prove that a test fails; nor should avoidable barriers be dismissed as a test of character.
When a student’s score differs sharply from their usual work, look at evidence across settings. Was the difficulty limited to the timing, did a particular question type cause errors, or did the independent knowledge itself remain fragile? A single explanation should not be imposed without investigation.
A five-minute diagnostic after returning a test
- Find one representative error. Choose a question that reveals a repeated pattern, not necessarily the hardest question.
- Ask for the thinking. Let the student explain the original approach before showing a model answer.
- Locate the first unsupported step. Distinguish a calculation slip, a language misunderstanding and a missing concept.
- Teach that small point. Use one accurate explanation or contrasting example.
- Test a new case. Remove the worked example and see whether the learner can solve independently.
Some students need more than five minutes. The routine is a starting check, not a full diagnostic evaluation. Its purpose is to turn the paper into evidence that guides teaching.
Three complementary ways to check learning
- Exam-style questions: useful for rehearsing the knowledge and performance demanded by a particular assessment.
- Short explanations: useful for revealing whether the learner understands why a method or claim works.
- Changed problems or delayed checks: useful for seeing whether knowledge survives new wording, a different context or a gap in time.
These approaches can support one another rather than compete. The right mix depends on the subject and learning stage. A practical laboratory skill, for example, cannot be assessed solely by answering questions about laboratory equipment.
Singapore MOE’s 2026 G1 Computing syllabus explicitly describes assessment as integral to learning and calls for feedback that moves learning forward. It also distinguishes assessment for learning and assessment of learning, with skill-building tasks, projects and timed tasks serving different purposes.
What parents should ask after an unexpected score
- Which specific learning outcome was the question intended to assess?
- Did the child choose the wrong method or use the right method inaccurately?
- Was the student able to explain the concept in a fresh, untimed setting?
- What will the next lesson or revision session do differently?
- How will we know the missing skill has been repaired independently?
An examination score can be meaningful without defining a child’s worth or ceiling. Treat it as evidence that deserves interpretation, not as an identity label.
What schools can measure besides the final grade
Alongside attainment, schools can consider patterns in misconceptions, consistency of responses to different tasks, improvement after feedback and the quality of work under meaningful conditions. It is also helpful to check whether an intervention actually addressed the difficulty it was introduced to solve.
The EEF’s Embedding Formative Assessment evaluation reported a positive average impact in participating English schools. The result concerns a specific programme and context, not an assurance that every assessment reform produces the same gains.
Frequently asked questions
Are examinations a poor way to measure learning?
Not necessarily. They can measure selected knowledge and skills effectively when aligned with a clear purpose and well designed. Problems arise when their evidence is incomplete, interpreted too broadly or disconnected from teaching.
Does a high grade guarantee deep understanding?
No. It demonstrates performance on the assessed tasks. Use explanation and changed questions to check broader transfer where it matters.
Why can two students with the same score need different support?
Because the underlying error patterns may differ. Inspect their work and reasoning to identify the different needs.
Should parents stop using past papers?
No. They are valuable for learning expectations and exam technique, but should be supplemented with conceptual explanation, feedback and independently chosen methods.
The final idea
Examinations fail when the score replaces the question it was meant to help answer: what does this learner know, what remains uncertain and what should happen next? The most useful assessment does not end with a number. It creates the possibility of better teaching and better learning.
Explore the other How Things Fail guides
- How Examinations Fail | When Scores Stop Measuring Understanding (this guide)
- How Decision-Making Breaks | When Bias Overrides Evidence
- How Innovation Stalls | When Good Ideas Cannot Scale
- How Early Warning Systems Fail | When Alerts Do Not Reach People
To connect these explanations with the wider series, see How Education Fails, How Strategy Fails and the unchanged Secondary 1 Mathematics Tutor Clementi | Small Groups Tutorials.
Evidence and further reading
- Singapore MOE: Primary Mathematics Syllabus — Assessment in the Classroom
- Singapore MOE: G1 Computing (2026), School-Based Assessment
- UNESCO: Assessment for Improved Learning Outcomes
- Education Endowment Foundation: Teacher Feedback to Improve Pupil Learning
- EEF: Embedding Formative Assessment
- eduKateSG: How Mathematics Learning Stalls
The situations above are illustrative teaching examples, not claims about identifiable students or institutions. Good responses depend on the actual evidence, duties and circumstances.