Why quizzes reveal gaps better than rereading

A quiz replaces neither understanding the lesson nor complex exercises. Its value comes above all from the loop it creates: answer, identify the mistake, correct it, then retest later whatever was still shaky.

In private tutoring, the gap between rereading and self-testing is very concrete. A student can reread a revision sheet twice and recognise every concept. As soon as you close the material and ask “which property would you use here?”, some certainties vanish. That is not a failure of the quiz: it is exactly the information you are looking for.

1. The problem of the illusion of mastery

Rereading is useful for discovering content, checking a definition or going back over an explanation. It becomes less informative when it is used to answer the question “do I know this?”. As long as the answer stays in front of them, the student can mistake familiarity for the ability to retrieve the information alone.

A quiz changes this situation. The question appears without the solution and forces a decision. Even an “I don’t know” provides clearer data than a third reading of the paragraph.

Research on retrieval practice has described this mechanism for a long time. A systematic review by Agarwal, Nunes and Blunt coded 50 experiments carried out in schools and classrooms, with a total of 5,374 participants. Of the 49 effect sizes reported, 57% were medium or large in favour of retrieval practice. The authors observed benefits across different school levels and subjects, while noting that only 6% of the experiments came from non-Western industrialised countries.

A quiz should reveal something that reading does not show

After a chapter, compare these two situations:

• “When I reread the definition, everything seems clear to me.”
• “Without looking, I can give two elements out of three, but I always forget the condition for applying it.”

The second statement makes it possible to act. The student knows exactly what they need to review, and the tutor can decide whether the problem comes from memorisation, a confusion or understanding that is still incomplete.

2. A test does not only assess learning: it can also contribute to it

A quiz is not just a measurement placed after revision. Searching memory for an answer can itself become a learning episode. This is often called the testing effect (in French, the effet de test).

Christopher Rowland’s meta-analysis compared the effects of retrieval tests and restudying on retention. It concludes that testing has an overall benefit and notably reports larger effects when the initial test requires recalling an answer rather than simply recognising it.

An experiment by Karpicke and Blunt published in Science also found that university students learning scientific texts performed better at a delayed test after retrieval practice than after several study conditions, including on comprehension and inference questions.

These results do not mean that a ten-question multiple-choice quiz is enough to learn any chapter. A quiz on definitions does not automatically measure the ability to write an essay or solve a new problem. The questions need to be aligned with the skill you are aiming for.

A good quiz produces effort, then a correction

If the question is so easy that the answer comes without any thought, it provides little diagnosis. If it is so difficult that the student guesses everything, it provides little learning. The useful level lies in between: accessible enough to attempt retrieval, demanding enough to distinguish what is solid from what is not yet.

3. Mixing easy, intermediate and difficult questions

A quiz made up entirely of basic questions can give a reassuring score without checking mastery of the chapter. Conversely, a quiz built only from trick questions or very complex applications can suggest that a student knows nothing when they actually have the basics.

A simple structure is to vary three levels:

Level	What the question checks	Example
Foundation	Directly retrievable knowledge	Give a definition, a formula or a date.
Intermediate	A distinction or link between two pieces of knowledge	Explain why two concepts should not be confused.
Application	Using knowledge in a slightly different situation	Choose the relevant method in a new example.

This is neither an official classification nor a universal marking scheme. It is a practical way to avoid a quiz made up solely of small recognition questions.

4. Analysing wrong answers instead of just counting points

The most useful outcome of a quiz is often not the final percentage, but the list of mistakes and hesitations. After each wrong answer, quickly classify the likely cause:

• forgetting: the information had been understood but is no longer accessible;
• confusion: two concepts are mixed up;
• shaky understanding: the definition is known but not the mechanism;
• application: the knowledge is available but the student does not know when to use it;
• reading: the question was misinterpreted.

This classification is not an automatic scientific diagnosis. It serves to decide on the next action. Forgetting may lead to another recall. A confusion calls for comparing the two concepts. An application problem calls for an exercise instead.

The wrong options in a multiple-choice quiz can themselves create confusion

A multiple-choice quiz presents several answers, some of which are wrong. If the student selects a wrong option and then receives no correction, they may hold on to that incorrect information.

In a study by Butler and Roediger, university students worked on texts and then took multiple-choice tests with immediate feedback, delayed feedback or no feedback. Both feedback conditions improved correct answers and reduced later reuse of the wrong options compared with no feedback. The context is university-level, but the practical message is clear: a revision quiz must allow the student to find the right answer and understand the mistake.

5. When should you redo a quiz?

Immediately redoing exactly the same questions can check that the correction has just been understood, but it is not solid proof of lasting memorisation. The student may simply be holding the answer in very short-term memory, hence the value of testing the student after a delay.

It is useful to combine two moments:

1. correct now: understand why the answer was wrong and be able to give the right one;
2. retest later: check that the information remains accessible once the correction is no longer fresh.

The systematic review by Agarwal and colleagues found benefits of retrieval practice with different delays and different feedback timings. The synthesis work of Roediger and Butler also indicates that repeated, spaced retrieval accompanied by feedback is a particularly useful combination for retention.

There is no single frequency to impose on every chapter. In practice, you can retest mistakes after the correction, then put them back into a quiz a few days later. Knowledge that stays solid can come back less often; knowledge that fades needs to be reviewed sooner.

This logic follows directly from active recall: try to retrieve, check, correct, then start again after a delay.

6. Creating a useful quiz from a chapter

A good quiz does not try to turn every line of the lesson into a question. It selects what the student genuinely needs to be able to retrieve or use.

Start by identifying:

1. the essential definitions and reference points;
2. the distinctions that often cause confusion;
3. cause-and-effect relationships or mechanisms;
4. the methods that need to be recognised;
5. the typical mistakes in the chapter.

Then turn these elements into varied questions.

Objective	Example question
Definition	What is an affine function?
Distinction	What is the difference between an image and a preimage?
Explanation	Why can a rise in the policy interest rate slow down demand?
Application	Which property solves this situation, and why?
Common mistake	Why is this equality incorrect?

How many questions? Enough to cover the objective, not to inflate the score

There is no universal optimal number of questions per chapter. Ten questions may be enough to check a very targeted set of knowledge; they will not be enough to cover a dense chapter involving several skills.

The best criterion is coverage: has every important concept been tested? Are the questions different enough to avoid a good score resting solely on a series of easy definitions? The quiz should stay short enough for the student to analyse their mistakes afterwards.

Flashcards meet a slightly different need: they make it possible to repeat small units of knowledge over time. A quiz can mix several concepts from the same chapter and give a broader snapshot at a given moment.

7. Using the results to prepare the next session

A quiz becomes genuinely useful when it changes what happens next. Suppose a student scores 8/10. The two wrong answers both concern the same concept. The decision is not necessarily “redo the quiz”: it may be more relevant to go back over this concept with the tutor and then test it in a different exercise.

Before the next session, three pieces of information are often enough:

• the questions that were missed;
• the answers given or the hesitations;
• what the student understood after the correction.

The tutor can then start by checking the weak point instead of going over the whole chapter again.

This logic ties in with the broader follow-up described in our article on measuring a student’s progress in private tutoring: a score is only truly informative when you know which skill it represents, which mistakes make it up and what the student can then redo without help.

A concrete example of the quiz → session → new quiz loop

A student revising functions gets all the definition questions right, but still confuses image and preimage in three application situations. Instead of having them reread the whole chapter, you target this distinction. They rephrase the two concepts, work through two exercises with the tutor, then tackle a new question without help.

A few days later, the quiz should not simply repeat the three original questions word for word. A new wording makes it possible to check whether the distinction has become usable and not merely memorised in the previous order.

Should the correction be shown immediately?

A revision quiz must provide a correction, but research does not justify a universal rule that every correction should always appear the very second the student answers.

Kulik and Kulik’s meta-analysis on feedback timing already showed different results depending on the situation: applied studies using real quizzes and course content generally favoured immediate feedback, while some experimental paradigms obtained other results.

The practical rule can therefore be simpler: first let the student genuinely try to answer, then make sure they get the right answer early enough to understand and correct their mistake. In a multiple-choice quiz, this correction is particularly important to avoid holding on to a wrong option.

Using a Learn pack quiz as a diagnosis, not just a score

On NOEO, the student can, on their own initiative, create a Learn pack from a completed session, from several completed sessions with the same tutor in the same subject, or from content they provide themselves. A Learn pack contains a revision sheet, flashcards and a quiz.

The quiz can be used to check what is still available after the session. Its value is not just in getting a number of right answers: the mistakes show which concepts to go back over in the revision sheet, with the flashcards or in the next piece of work.

The Learn pack is not created automatically after each session, and the tutor does not create it in the student’s place. Exercises sent by the tutor in the Session report remain separate from the Learn pack.

A simple way to use it, then, is to read or review the concept, close the material, take the quiz, analyse each mistake, then retest the weak points later. If a wrong answer mainly reveals an application problem, the right follow-up is not another multiple-choice quiz: it is an exercise.