How to teach through mistakes: a practical technique for online courses

In most online courses, a bug has a very short life. The student chooses the wrong answer, sees a red cross, clicks "try again" and moves on. The teacher gets the statistics of wrong answers, but rarely knows why people got it wrong. As a result, the course seems to test knowledge, but loses one of the most valuable opportunities - to teach on the wrong solution.

An error can be useful if it can be safely investigated. Not every failure automatically teaches: a repeated mistake without explanation can only reinforce the wrong approach. Therefore, a well-thought-out cycle is needed: an attempt, analysis of the cause, correction of the model, a new attempt and verification in a different situation.

In educational design, it is important to create a space where the student is not afraid to show how he thought. It is the process of reasoning that gives the teacher material to help, and the student a chance to understand his typical pitfalls.

Not all bugs are created equal

Suppose two students answered a financial literacy question incorrectly. The first does not know what compound interest is. The second knows the rule, but misread the condition. If both are shown the same formula explanation, one will get the help they need and the other almost none.

At least four types should be distinguished: lack of knowledge, incorrect application of a known rule, error of execution, and error of interpretation of the result. This is not a universal scientific classification, but a practical tool for the author of the course.

In the sales course, there can also be various reasons for the error: the manager does not know the product, did not hear the customer's question, missed a critical stage of the dialogue, or jumped to a conclusion too quickly. If the feedback is limited to "wrong answer", the reason will remain invisible.

When to show the correct answer

The common idea is that the student must "go it alone" at all costs. But if a person has no basic support at all, endless failed attempts can only increase the confusion. In such a situation, a brief explanation or demonstration of one step is appropriate.

Another case is that the student is already close to the correct solution. Here, a ready-made answer can deprive useful practice. It is better to ask a leading question: which fact in the problem did you not use? Does this formula match the units? What will happen to the result in the limiting case?

So the prompt needs to be dosed. First point out the problem, then help to find the correct principle, and only if it did not work - show a completed example. After the demonstration, give a similar new task so that the student does not remain only a passive observer.

Attempted Explanation: When it makes sense

Sometimes it is helpful to ask the student to guess the answer before they have all the tools. For example, before a lecture on the theory of probability, you can show a life situation and ask to estimate how plausible a certain conclusion is.

The purpose of this exercise is not to have the participants guess the formula. It helps to manifest initial ideas. After the explanation, they can go back to the first version and compare exactly what has changed.

However, this format should be used with care. For a difficult or high-stakes topic where students have no foothold at all, it's best to give a basic framework first. Pedagogical reception is not an end in itself; its usefulness depends on the initial level and subsequent feedback.

What good feedback should look like

Let's compare the three messages after the error. First: "Wrong". Second: "The correct answer is B." Third: “You correctly identified the client's primary goal, but proposed an action before checking the contract's limitations. Go back to step two and specify the condition."

The third option points to a specific error and leaves the student to do some of the work. It does not devalue the entire attempt and does not turn the teacher into a ready-made answer machine.

Useful feedback answers three questions: what exactly didn't work, why it's important, and what to do next. Excessive praise or a long lecture after every wrong click is often only a distraction. Accuracy and the ability to immediately repeat the action are important.

Case study: Help desk training on false dialogs

Let's imagine a course for operators who work with order returns. The student sees three short dialogues. In one, an employee promises a client a result without checking the status. Otherwise, he uses the correct rules, but speaks harshly. In the third, he asks a lot of questions, but does not agree on the next step.

The contestant must identify where exactly the error occurred, what risk it created, and what one line should have been said differently. Then he gets a role-playing situation where he has to conduct a short conversation himself.

A strong exercise isn't all about finding a "bad" phrase. She shows that different mistakes have different consequences: a wrong promise can cause financial losses, a harsh tone can lead to conflict, a lack of next step can lead to a repeat appeal.

This is an educational example, not a case study of a specific company. Its value is that it demonstrates the difference between a memory test and the exercise of professional judgment.

Safe errors and dangerous errors

Not every mistake can be allowed in real work. In medicine, industrial security or financial transactions, an incorrect decision can create a serious risk. That is why simulations, training environments and gradual admission to independent action are especially important in such areas.

For example, an employee can practice a process in a test CRM where a wrong step will not change the actual customer data. A medical student — on a scenario with a conditional patient. The future manager is in a role-playing conversation, where a wrong phrase will not harm the real team.

The main thing is to honestly mark the boundaries of the simulation. A conditional situation does not reproduce all the emotions and consequences of real work, so after it you may need practice under the supervision of a mentor.

How to turn typos into an exercise library

Each repeated type of wrong decision can become a learning material. To do this, you need to collect impersonal examples and describe not the person who made a mistake, but the situation and the reason.

For example, in an analytics course, students regularly confuse percentage gain with percentage point difference. A course writer can create a short exercise with two graphs and several plausible misinterpretations. After making a choice, the student should explain what is wrong, not just see the correct answer.

Such a library is useful to review quarterly. If, after editing a lesson, a certain error occurs less often, this is a signal that the change helped. If the frequency does not decrease, it is worth checking not only the explanation, but also whether the exercise corresponds to the desired skill.

Why students need remediation

If every first mistake irreversibly lowers the grade, students quickly become risk averse. They choose the easiest tasks, look for hints before trying, and are less willing to show incomplete understanding.

For training exercises, it is useful to allow a second attempt after a brief feedback. But the second attempt should not be a mechanical clicking of another answer. Ask to briefly explain what has changed in reasoning. If the student can't do this, clicking correctly doesn't mean mastery.

In the final assessment, the rules may be stricter. It is important to clearly separate the space of practice from the space of certification so that students understand where they can experiment and where they need to demonstrate a stable result.

Parsing a mistake in a group without humiliating the member

Public analysis of a mistake can be useful if it focuses on the decision rather than the individual. It is better to show an impersonal fragment of the answer and ask: "What assumption here could have led the author astray?". This opens up the discussion without turning it into an assessment of a particular person.

The teacher can show an example of his own false hypothesis and explain how he tested it. So students see that revising a position after new data is a sign of professional work, not weakness.

Set rules separately: don't mock answers, criticize arguments, not people, suggest a way to check. These simple conventions are especially important in online groups, where a short comment can easily be taken more harshly than intended.

AI as a bad decision simulator

Generative AI can help create variants of common mistakes for training exercises. For example, ask the model to generate a short incorrect solution with one plausible error. The student must find it and explain the reason.

But the AI ​​also makes mistakes, sometimes uncontrollably. The author must check whether the example has exactly one error, whether there are no additional contradictions, and whether the exercise corresponds to the level of the group. Do not automatically publish generated materials.

In the WeStudy LMS platform, learning materials and tasks can be organized, and the technical capabilities of a specific test format should be confirmed in the platform documentation. The pedagogical rule is unchanged: the result of the exercise should be a better professional decision, and not a quick guess of the correct option.

How to evaluate the effectiveness of error learning

Do not limit yourself to the percentage of correct answers on the second attempt. It can increase simply because the student remembered the right button. A much stronger indicator is whether the corrected rule is carried over to the new task.

Create parallel versions of cases. If the student made a mistake when calculating the percentage increase, next time give a similar problem with a different context. If he correctly explains why the approach has changed, there is reason to talk about progress.

A simple table is useful for the teacher: type of error, its frequency before analysis, frequency in the new problem, additional support that helped. The data should not become a ranking of the "worst students". Their purpose is to make the course more accurate.

Plan to retool one test per week

On the first day, choose the ten questions that students get wrong the most. Second, figure out what action each question is testing. Third, for half of the questions, create short explanations of the reasons for the typical wrong answers. Fourth, add a second attempt with a slightly changed condition.

On the fifth day, test the material on several volunteers. Ask them to comment on what they think before choosing. On the sixth day, correct ambiguous wording. Seventh - Publish the update as a draft or pilot to a separate group.

No need to change the entire course at once. One quality block with feedback can become a model for subsequent topics. After the pilot, compare not only test scores, but also students' ability to solve new problems without prompts.

How to build a "Museum of Errors" in a professional course

Some teachers collect only the best work of students to display as a sample. However, the second archive — the library of typical wrong decisions — can be useful for professional training. It is important that it consists of anonymized materials and does not allow identification of a specific participant.

For example, a marketing analytics course might create a collection of graphs with erroneous labels, incorrect axes, and overly bold causal inferences. For each example, add not a ready-made explanation, but three questions: what is asserted here, what data is missing, and what conclusion can be drawn more cautiously. After an independent attempt, the student sees the author's comment.

For future teachers, such a library may contain tasks with unclear instructions and grading rubrics that give different results to different examiners. For managers - sample letters where formally polite text leaves the client without a specific next step. Every discipline has its professional set of mistakes.

Update the collection not for quantity. If a new type of difficulty appeared in the educational process, add one high-quality example. If the old trap no longer corresponds to real practice, remove it. Such a museum becomes not entertainment, but a living reflection of how professional judgment develops.

How to avoid auto-guessing in tests

In a four-choice test, the student has the opportunity to find the correct answer at random. If the system allows an unlimited number of attempts without changing the condition, a person can go through the material almost without thinking. This is especially noticeable when after each click you can see which option was wrong.

One way to improve is to ask for a brief explanation of the choice before showing correctness. Another is to give a second question on the same principle, but with new numbers or context. The third is to create practical cases where the student must not only name the solution, but also justify it.

You don't have to turn every question into a long essay. For example, after choosing the correct formula, it is enough to ask for one sentence: "I chose this method because...". This small effort prompts awareness, and the teacher can see if random guessing is lurking.

For high-stakes tests, the rules should be stricter. But the training block can be built so that a mistake provides information for the next exercise, and not just lowers the score.

Personal error map instead of abstract score

A useful tool for the student is a small journal, where he records not every mistake, but those that are repeated. Four fields are enough: the situation, my initial assumption, what turned out to be wrong, how I will check myself next time.

For example, a student is studying SQL and regularly forgets that an INNER JOIN can change the number of rows in the summary. He writes not "again made a mistake in JOIN", but more specifically: "before aggregation, I check the uniqueness of the key and the number of rows after the connection". Next time, this check becomes his own operating rule.

For humanities education, a map can contain other types of pitfalls: mixing fact and interpretation, overgeneralizing from a single example, and lacking a counterargument. In each case, it is important to name the action that will help to avoid repetition.

Check the log every two weeks. If a certain error no longer appears in new tasks, remove it from the active list. If the same reason returns despite repeated explanations, other pedagogical support or a return to the basic skill may be needed.

Teacher's role during the trial: judge or researcher

When a student makes a mistake, the teacher can automatically switch to the evaluation mode: right or wrong, how many points, what should have been remembered. But in a learning situation, it is sometimes more useful to act as a researcher. To ask what logic the person used, what seemed to him the most important thing in the condition, why he rejected another approach.

Such research does not mean that all answers are equally good. On the contrary, it helps to more precisely explain the limits of the correct decision. A student may have a valid idea, but not apply it there; another is to accidentally name the correct option without understanding the reason.

How the teacher talks about his own mistakes is also important. The phrase "I also misread this table at first, so now I always check the units" shows a professional way of dealing with uncertainty. This is not a demonstration of weakness, but a model of behavior that the student will be able to transfer to his profession.

What to say to a student after the third failed attempt

The third error in a row is the moment when a short message is no longer enough. Do not answer "try again" without new information. Ask the student to describe what they think is constant about the task and what step is questionable. If the answer is unclear, go back to a simpler example that shows the same principle.

Sometimes the best way out is to take a short break. After several unsuccessful attempts, a person can continue to press the options automatically. Offer to return to a similar task with a new context after a certain time. If the problem is systemic, it is worth checking the basic prerequisites, and not increasing the number of exercises of the same type.

Good maintenance doesn't mean endlessly sticking to one task. The teacher has the right to change the level of difficulty, show part of the solution or suggest an alternative way. The main thing is to return to the student the possibility of independent action after the help, and not to complete the entire process for him. For group courses, it's helpful to keep a short question channel: the student can report exactly which prompt worked, and the content author can see if the explanation needs to be changed for the next stream. This turns even a failed attempt into information to improve the course and help better teach subsequent students.

How to handle errors in open answers

Multiple-choice tests are easy to analyze automatically, but open-ended solutions are important in professional training. For example, a student should write a short answer to a client or justify the choice of a research method. Here the error may not be one, but consist of several interrelated assumptions.

When checking, first find the major gap between what the task required and what the student did. You don't need to fix all the text in red. Give one or two priority recommendations that can be applied in the next attempt. If the deficiency is related to argumentation, ask to add proof. If with the logic of actions, explain the order of selection.

Leave the option to show the patched version and briefly describe what changed. So the teacher sees not only the mistake, but the learning process. In the next task, you can check whether the student has transferred this principle to a new context.

Agree on which mistakes are part of the training and which need to be stopped and helped by the instructor. Use safe fictional data and retries for the tutorial. After a correction, ask for a brief explanation of the reason for changing the decision: this will help distinguish a meaningful correction from a random guess of the next option.

Conclusion

Learning by making mistakes is not about romanticizing failure or requiring you to experiment without rules all the time. It's a thoughtful opportunity to see why a certain solution didn't work, fix it, and test a new approach.

A strong online course creates a safe space for experimentation, provides specific feedback, and does not penalize learning errors where they are part of practice. LMS platform WeStudy or other digital environment can support the organization of these steps, but quality pedagogical design remains crucial.

Sources

1. Education Endowment Foundation — Metacognition and Self-Regulated Learning.

2. Education Endowment Foundation — Feedback.

3. The Learning Scientists — Retrieval Practice.

4. Kapur — Productive Failure in Learning.