Why the Same Student Gets Different Exam Scores: What Test Formats Really Measure

A student receives 90% on a multiple-choice examination, 65% on a written test, and 75% during an oral assessment.

The subject is the same.

The student is the same.

The underlying knowledge has not suddenly disappeared between examinations.

So which score represents the student’s actual ability?

The highest?

The lowest?

The average?

Or is the question itself misleading?

An examination result is not a direct measurement of everything a person knows. It is an estimate based on particular tasks, performed under particular conditions, and evaluated according to particular criteria.

Change those conditions, and the result may change.

This does not automatically make examinations unreliable or unfair.

But it raises a question that education cannot afford to ignore:

When different examination formats produce different scores, are they revealing different abilities, or are they measuring the same ability with different degrees of accuracy?

The answer matters to students, parents, teachers, examination providers, and institutions that use test scores to make decisions.

1. An Exam Score Is Not the Same Thing as Knowledge

Imagine a student who understands the causes of the First World War.

In a multiple-choice test, the student correctly identifies relevant historical events, political alliances, and chronological relationships.

In a written examination, the same student must explain how those factors contributed to the escalation of the conflict.

During an oral examination, the student must defend an interpretation while responding to questions.

All three assessments concern historical knowledge.

But they do not require exactly the same performance.

The multiple-choice test may assess recognition, factual discrimination, and the ability to select an answer from alternatives.

The written examination may assess explanation, evidence selection, and argument construction.

The oral examination may additionally assess spontaneous reasoning, listening, and the ability to respond to follow-up questions.

A student can be strong in one combination of skills and less developed in another.

That difference is not necessarily a contradiction.

The mistake is assuming that every examination format measures an identical quantity called knowledge.

Knowledge is not always demonstrated through one action.

Recognising a fact, recalling it without assistance, explaining its significance, and applying it to an unfamiliar situation are related but distinguishable performances.

A meaningful assessment must identify which of them matters.

2. Multiple-Choice Tests: Recognition Is Not the Same as Independent Production

Consider a biology question:

Which organelle is primarily responsible for producing ATP during aerobic respiration in eukaryotic cells?

A. Ribosome
B. Mitochondrion
C. Golgi apparatus
D. Lysosome

A student selects B.

The answer is correct.

But what does this response establish?

It provides evidence that the student can identify the correct option among four alternatives.

It does not necessarily establish that the student could independently explain oxidative phosphorylation, the role of the electron transport chain, or the relationship between cellular respiration and ATP production.

Those would require additional evidence.

Now remove the options and ask:

Explain how mitochondria contribute to ATP production during aerobic respiration.

The student must retrieve relevant knowledge, organise it, and construct an explanation.

This task places different demands on the student.

A higher multiple-choice score is therefore not automatically evidence that the multiple-choice examination was easier.

The two tasks may assess different levels or aspects of competence.

Multiple-choice questions can assess complex reasoning

It would also be incorrect to assume that multiple-choice tests assess only memorisation.

A carefully designed question can require students to interpret data, evaluate an experiment, compare explanations, or apply a concept to an unfamiliar situation.

For example, students may be given a graph showing changes in enzyme activity and asked to identify the explanation most consistent with the evidence.

That requires more than recognising a definition.

The quality of the question matters as much as its format.

Nevertheless, selected-response questions and independently constructed responses provide different kinds of evidence.

The distinction is examined from another perspective in Why Multiple-Choice Tests Trick Even Good Students.

The important question is not whether multiple-choice tests are good or bad.

It is whether the specific questions provide appropriate evidence of the intended competence.

3. Written Exams: The Student Must Build the Answer

A written examination removes many of the cues available in multiple-choice tasks.

Students may need to retrieve facts without prompts, select relevant information, organise arguments, and express relationships accurately.

Consider a mathematics question.

Solve the equation and explain why your method is valid.

A student who can recognise the correct solution among four alternatives may still struggle to produce the steps independently.

Another student may calculate correctly but provide an incomplete justification.

A third may understand the procedure but make a transcription error.

These students have different performance profiles.

The written task makes some differences visible that a multiple-choice task may not reveal.

But it also introduces additional demands.

Students must decide what to include, how much detail to provide, and how to make their reasoning understandable to another person.

These demands may be part of the assessment objective.

If the examination is intended to evaluate mathematical argumentation, explaining the method is essential.

If it is intended to assess only the ability to perform a calculation, extensive prose may be unnecessary.

A well-designed assessment should make that distinction explicit.

The earlier article When a Student Knows the Answer but Cannot Express It: What Is the Exam Actually Measuring? explores why the ability to express knowledge and the possession of knowledge should not automatically be treated as identical.

The format of the assessment determines how these abilities interact.

4. Oral Exams: Understanding Under Interaction

An oral examination introduces another dimension.

The student is no longer responding only to a written task.

The student may also be responding to a person.

Consider a German-language examination.

The examiner asks:

Was würden Sie tun, wenn Sie Ihren Zug verpassen würden?

The student must understand the question, retrieve appropriate vocabulary, construct a hypothetical response, and speak under time pressure.

A written response might allow more time for revision.

An oral response requires immediate production.

The examiner may ask a follow-up question:

Warum würden Sie sich so entscheiden?

Now the student must justify the decision.

This is not simply the written examination spoken aloud.

It is an interactive task.

The ability to maintain communication, respond appropriately, repair misunderstandings, and develop an answer may be relevant to the assessment.

The examiner’s role matters

In an oral examination, the examiner’s behaviour can influence the interaction.

The wording of follow-up questions, the time allowed for responses, and the consistency of administration may affect what students have an opportunity to demonstrate.

Well-designed oral examinations therefore need clear procedures and appropriate examiner training.

Standardisation does not mean that every conversation must sound mechanically identical.

It means that candidates should have sufficiently comparable opportunities to demonstrate the assessed abilities.

A student who performs poorly in an oral examination may need more speaking practice.

But the result alone does not establish that the student lacks grammatical knowledge, reading comprehension, or the ability to write accurately.

It provides evidence about performance in the assessed oral tasks.

5. Why One Student May Excel in Writing but Struggle in Speaking

Some students can organise complex arguments effectively when they have time to plan.

They may produce excellent essays, precise explanations, and well-structured written responses.

But spontaneous speaking requires additional processes.

The student must retrieve language quickly.

They must monitor their own speech while continuing to communicate.

They may need to respond before they have fully formulated the sentence.

In an interactive examination, they must also process another person’s questions.

A student may therefore possess substantial knowledge while struggling to retrieve and organise it quickly enough for oral performance.

The reverse is also possible.

A student may communicate confidently in conversation but struggle to produce formal written arguments.

The student can explain an idea through examples, gestures, and spontaneous reformulation.

Yet a written examination may require a precise structure, explicit logical connections, and an appropriate academic register.

Neither profile automatically proves that one student is more intelligent or knowledgeable than the other.

The profiles show that the students perform differently across tasks.

Whether those differences should affect the final grade depends on what the qualification is intended to certify.

6. Reliability: Would the Result Remain Similar Under Comparable Conditions?

In assessment, reliability concerns the consistency of measurement.

Suppose a student takes two comparable tests designed to assess the same competence.

If the results differ dramatically, we need to investigate why.

Perhaps one test contains substantially harder questions.

Perhaps the student was ill.

Perhaps the marking criteria were applied inconsistently.

Perhaps the two tests did not actually assess the same abilities.

A reliable assessment aims to reduce variation caused by factors unrelated to the intended measurement.

But reliability does not mean that a student must receive exactly the same score every time.

Human performance varies.

Tasks vary.

Even well-designed assessments contain measurement uncertainty.

The important issue is whether the scores are sufficiently consistent for the decisions being made.

A numerical example

Imagine a student receives:

• 78% on one mathematics test;
• 81% on another comparable mathematics test;
• 76% on a third.

These results are reasonably close.

Now imagine the scores are:

• 92%;
• 48%;
• 85%.

The larger variation raises questions.

Were the tests comparable?

Did they assess the same content?

Were the conditions similar?

Was one test unusually sensitive to a particular weakness?

Without investigating those questions, we cannot confidently interpret the differences.

We should not automatically conclude that the student’s knowledge changed dramatically.

Nor should we assume that the tests were defective.

The evidence is insufficient to establish either explanation.

7. Validity: Is the Exam Measuring What It Claims to Measure?

Reliability is not the same as validity.

An assessment can produce highly consistent scores while failing to support the interpretation or use for which those scores are intended.

Imagine a test that asks students to memorise and reproduce fifty scientific definitions.

Students who study the definitions may achieve consistently high scores.

The test might reliably measure their ability to reproduce those definitions.

But could we use the result to conclude that the students can design experiments, interpret evidence, or apply scientific principles?

Not without additional justification.

Those are different competencies.

A score is not valid or invalid in isolation.

The important question is whether the evidence supports the conclusions and decisions made from it.

The problem of overinterpretation

A student receives a high score on a grammar test.

Someone concludes:

“This student can communicate confidently in German.”

That conclusion may be unsupported.

The grammar test may provide strong evidence of grammatical knowledge while providing limited evidence of spontaneous speaking ability.

Similarly, a student performs well in an oral examination.

Someone concludes:

“This student can write accurate academic essays.”

Again, the evidence may be insufficient.

The examination result has been extended beyond the abilities directly assessed.

This is one reason why passing a language examination does not automatically establish complete practical language competence.

A qualification may provide valuable evidence.

But that evidence must be interpreted within the scope of the assessment.

8. A Reliable Test Can Still Be Too Narrow

Consider a history examination containing one hundred factual questions.

Suppose the questions are carefully written, marking is objective, and scores are highly consistent.

The examination may reliably assess factual knowledge.

But if the educational objective includes interpreting historical sources and constructing arguments, the assessment covers only part of the intended domain.

This is a problem of coverage.

The test may omit important aspects of competence.

The opposite problem is also possible.

A written history examination may require lengthy essays even when the main objective is to assess factual understanding.

Students with weaker academic writing skills may receive lower scores despite possessing relevant historical knowledge.

In that case, the assessment may introduce additional demands that complicate interpretation of the result.

Neither problem can be solved simply by making the test longer.

A longer examination is not automatically a better examination.

The tasks must represent the abilities the assessment is supposed to measure.

9. Why Combining Different Formats Can Improve Assessment

If different formats reveal different aspects of performance, combining them can provide a more complete picture.

For example, a language assessment might include reading, listening, writing, and speaking.

Each component provides evidence about a different dimension of language use.

A science course might combine written explanations, calculations, data interpretation, and practical work.

A history course might combine factual questions, source analysis, and extended argumentation.

This approach can reduce dependence on one narrow type of task.

But combining formats introduces new decisions.

How much should each component contribute to the final result?

Should strong performance in one component compensate for weak performance in another?

Does the qualification require minimum competence in every component?

These are not merely mathematical questions.

They concern the meaning of the qualification.

When averaging can hide important weaknesses

Imagine a language student receives:

The arithmetic average is 72.5%.

But does that mean the student has satisfactory overall language proficiency?

The answer depends on the purpose of the assessment.

For a qualification requiring independent oral communication, the low speaking score may be important.

For a task involving primarily written document comprehension, the same profile may be interpreted differently.

Some formal examinations impose component-specific minimum requirements.

Others permit compensation between components.

Students should consult the actual rules of the examination rather than assuming that an average score tells the whole story.

10. The Same Format Can Contain Very Different Tasks

It is tempting to divide examinations into simple categories:

Multiple-choice tests measure recognition.

Written exams measure explanation.

Oral exams measure speaking.

But these categories are too broad.

A multiple-choice task may require complex analysis.

A written examination may consist entirely of short factual responses.

An oral examination may ask students to reproduce memorised material.

A practical assessment may require sophisticated reasoning or only routine procedural execution.

Therefore, the format alone does not determine the intellectual quality of an examination.

We must also examine the tasks.

What must students actually do?

Which information is available?

What reasoning is required?

How is the response evaluated?

Two written examinations can assess very different competencies.

Two oral examinations can differ dramatically in cognitive demand.

The relevant unit of analysis is not simply the format.

It is the relationship between the task, the intended competence, and the evidence produced.

11. When the Question Itself Changes the Result

Even within the same format, wording matters.

Consider a mathematics question:

Find the value of x.

Now compare:

Determine the value of x and justify each transformation used in your solution.

The mathematical problem may be identical.

But the second instruction requires additional evidence.

A student may obtain full credit on the first task and lose marks on the second.

That does not necessarily indicate inconsistent marking.

The tasks have different requirements.

However, poorly worded questions can introduce uncertainty about those requirements.

If a student cannot determine what is being asked, the examination may partly measure the ability to guess the question writer’s intention.

The distinction is explored in When Exam Questions Have More Than One Meaning: Who Is Responsible for the Wrong Answer?.

The format and the wording work together.

Neither can be evaluated independently of the assessment objective.

12. Multilingual Education: When Language Becomes Part of the Score

Now imagine a student studying mathematics in German.

The student understands algebra and can solve equations accurately.

But a written examination uses unfamiliar academic terminology.

For example:

Begründen Sie, warum die Gleichung keine reelle Lösung besitzt.

The student must explain why the equation has no real solution.

This requires mathematical knowledge.

It also requires understanding the German instruction and producing an appropriate explanation.

If the student cannot interpret begründen, the result may underestimate their mathematical competence.

But if the educational objective explicitly includes explaining mathematical reasoning in German, the language demand is part of what must be learned.

The same distinction applies to biology, chemistry, physics, geography, and history.

A student learning mathematics in German through Fachsprache must develop both conceptual understanding and the specialised language used to demonstrate it.

A student studying biology in English faces a similar interaction between subject knowledge and academic communication.

These challenges cannot always be separated cleanly.

But recognising their different contributions can improve teaching and assessment.

The risk of treating every difficulty as a language problem

There is also a danger in the opposite direction.

A student may speak German imperfectly and struggle with mathematics.

A teacher might attribute every mistake to language.

But some errors may reveal genuine conceptual misunderstandings.

The student may misunderstand fractions, proportional reasoning, or the meaning of an equation.

Providing a translation will not necessarily solve those problems.

The correct intervention depends on identifying the source of difficulty.

This is why integrated language-and-subject education should not become subject instruction replaced by vocabulary exercises.

Both dimensions require attention.

13. What About Exam Anxiety and Time Pressure?

Performance conditions can influence scores.

A student may know the material but struggle under time pressure.

Another may perform confidently during a written test but become hesitant when questioned orally.

These differences can be relevant.

However, they should not automatically be interpreted as evidence that the examination is unfair.

Some tasks legitimately require performance under time constraints.

A professional qualification may need to establish whether a person can make accurate decisions within a limited period.

An oral language examination may legitimately assess spontaneous communication.

The important question is whether the performance conditions correspond to the intended competence.

If speed is irrelevant to the learning objective, excessive time pressure may introduce unwanted variation.

If rapid retrieval is essential to the competence, timing may be justified.

Similarly, anxiety can affect performance, but a low score should not automatically be attributed to anxiety without supporting evidence.

Students benefit from identifying whether their difficulties concern knowledge, task familiarity, time allocation, or performance under pressure.

Those are different problems requiring different forms of preparation.

14. Fairness Does Not Mean Giving Every Student the Same Score

A fair assessment is not one in which everyone performs equally well.

Students genuinely differ in knowledge, preparation, skills, and performance.

The purpose of assessment is often to identify relevant differences.

Fairness concerns whether those differences are evaluated appropriately.

Consider two students taking an oral examination.

One speaks fluently and responds accurately.

The other understands the topic but cannot maintain the required interaction.

If the examination assesses oral communication, different scores may be justified.

Now consider a mathematics examination in which students are required to explain simple calculations through unnecessarily complicated language unrelated to the mathematical objective.

If the wording creates avoidable barriers, the resulting differences may be harder to interpret as differences in mathematical competence.

Fairness therefore depends on the relationship between the task and the ability being assessed.

It does not require removing every challenge.

It requires making the challenges educationally relevant.

15. Can Two Different Scores Both Be Correct?

Yes.

This is one of the most important conclusions.

Imagine a student receives 88% on a test of grammatical recognition and 62% on an assessment of spontaneous speaking.

Both scores may be accurate descriptions of performance on their respective tasks.

There is no contradiction if the assessments measure different competencies.

The problem appears when someone treats the scores as competing estimates of one identical ability.

The same principle applies to academic subjects.

A student may perform strongly in factual recall and less strongly in constructing arguments.

Another may solve practical problems successfully but struggle to explain the theoretical principles involved.

Different scores can reveal a meaningful profile.

But if two assessments genuinely claim to measure the same competence under comparable conditions, large score differences warrant investigation.

The distinction between different competencies and inconsistent measurement is essential.

Without it, discussions about examination fairness become confused.

16. How Students Should Interpret Different Exam Results

A student who receives different scores should not immediately conclude that one result is the true score and the others are mistakes.

Instead, the student should examine what each assessment required.

Did the multiple-choice test provide cues that helped with recognition?

Did the written examination require independent recall?

Did the oral examination demand rapid responses?

Did the questions require explanation, calculation, interpretation, or evaluation?

Were the marking criteria different?

Were the conditions comparable?

The answers can help identify a useful development plan.

A student who recognises correct grammar but cannot produce it needs practice in active retrieval and language production.

A student who knows historical facts but cannot construct an argument needs practice in explaining relationships and supporting claims.

A student who calculates accurately but struggles with mathematical explanations needs practice making reasoning visible.

A student who understands a subject in one language but not another may need specialised academic-language instruction.

The most useful result of assessment is not always the score itself.

It is the information the assessment provides about what to develop next.

17. What Teachers Should Learn from Conflicting Scores

When a student performs differently across formats, teachers have an opportunity to investigate the underlying competence.

The first step is to avoid reducing the difference to a personal characteristic.

Statements such as “This student is bad at exams” or “This student is only good at tests” explain very little.

A more productive approach compares the demands of the tasks.

What knowledge was required?

What response format was used?

What additional skills were involved?

Which criteria determined the score?

Where did the student lose marks?

Teachers can then use targeted tasks to test possible explanations.

If a student performs poorly in written science but explains the same concept accurately during discussion, the teacher can investigate whether the difficulty concerns written scientific communication.

If a student succeeds on familiar problems but fails when the context changes, the issue may concern transfer of knowledge.

If a student performs well when given answer options but struggles without them, independent retrieval may need development.

These observations can guide teaching.

They should not be used to invent certainty about a student’s internal abilities.

Assessment evidence remains partial.

18. What Schools and Examination Providers Should Consider

Institutions that use examination results for important decisions need to consider the intended interpretation of those results.

A qualification should communicate what successful candidates have demonstrated.

If a certificate claims broad communicative competence, the assessment should gather appropriate evidence across relevant language skills.

If a mathematics examination claims to assess problem-solving ability, its tasks should require more than reproducing memorised procedures.

If an academic programme expects students to construct arguments, assessment should include opportunities to demonstrate that ability.

Institutions must also consider the consistency of marking.

Extended written answers and oral performances may require carefully defined criteria, examiner training, and moderation.

Selected-response tests may offer more straightforward scoring but still require attention to question quality and content coverage.

No format eliminates the need for thoughtful assessment design.

And no single format automatically guarantees validity, reliability, or fairness.

19. The Three Educational Contexts

The meaning of a performance also depends on where it occurs.

In the learning process

A teacher can use several formats to investigate understanding.

An incorrect answer becomes an opportunity to identify a gap, ask follow-up questions, and adjust instruction.

The goal is development.

In a formal examination

The student must demonstrate competence through the authorised tasks and criteria.

The examiner cannot replace the published requirements with a personal judgement about what the student probably knows.

The goal is assessment.

In real life

Competence may be demonstrated through practical outcomes.

A person may communicate successfully despite grammatical imperfections.

A technician may solve a problem effectively without producing an academic essay.

A professional may explain a complex decision clearly in conversation but struggle with a particular written test format.

The goal is successful action within the relevant context.

These contexts overlap but are not interchangeable.

A good educational system should help students move between them without assuming that success in one automatically guarantees success in all three.

20. The International Perspective: Different Systems, Different Evidence

Students who move between educational systems often encounter unfamiliar examination formats.

One system may emphasise extended written explanations.

Another may use more selected-response questions.

A third may place greater weight on oral performance, coursework, practical tasks, or continuous assessment.

These differences can affect how students demonstrate their knowledge.

A student who previously achieved high scores may initially struggle in a new system without having lost the underlying subject knowledge.

The student may need to understand the new assessment expectations.

This is relevant to international learners supported through Levitin Language School and its American educational branch, Language Learnings.

Language learning, academic subject support, and learning subjects through another language represent connected but distinct educational needs.

A learner preparing for an English-language examination may benefit from understanding the relationship between proficiency levels and assessed skills, including the distinctions outlined in the English examination levels guide.

A learner studying German for academic purposes may need to connect linguistic competence with the subject-specific demands of their educational programme.

A learner preparing for school mathematics may need to develop mathematical reasoning and the ability to demonstrate that reasoning in the expected format.

Effective preparation begins by identifying what the examination actually requires.

Conclusion: The Same Student Does Not Always Produce the Same Evidence

Different examination scores do not necessarily reveal an inconsistency.

They may reveal different abilities.

They may reflect different task demands.

They may arise from ordinary measurement variation.

They may expose weaknesses in assessment design.

And sometimes they reveal that a qualification is being interpreted more broadly than its evidence permits.

The challenge is to distinguish these possibilities.

A multiple-choice score does not automatically establish independent explanatory ability.

A written examination does not automatically measure every dimension of practical competence.

An oral examination does not automatically reveal everything a student knows.

Each format provides a particular kind of evidence.

The educational responsibility is to understand what that evidence supports — and what it does not.

The same student can receive different exam scores without becoming a different learner. What changes is the evidence the examination asks them to produce.

A meaningful assessment system should recognise that distinction before turning a score into a judgement about a person.

Tymur Levitin
Founder & Director, Levitin Language School
Language Learnings — US Branch

Global Learning. Personal Approach.

© Tymur Levitin