Kai Kai gets a long algebra question wrong and writes “revise algebra” in his planner. The next evening he starts the chapter from the first example. He repeats several operations he can already perform, reaches the difficult question again and makes the same small transformation error. The revision was real work. It did not spend enough attention on the step that was failing.
Alicia does something similar in science. An explanation loses credit because the causal link is missing, so she rereads the entire topic. Tricia rewrites a complete essay after feedback identifies one weak connection between evidence and argument. Each learner treats the container around the error as the unit of repair.
A topic name is useful for organising a syllabus. It is often too broad for deciding what to do after one failed response. The educational question is whether the learner needs the whole topic rebuilt, one relationship retaught, one operation stabilised or a clearer way to complete the answer.
Alicia, Tricia and Kai Kai are fictional learners. Their examples are original teaching illustrations, not measured case studies. This article does not assume that every disappointing answer has only one cause. Its job is to show how to investigate the scope of a failure before committing the next revision session to an unnecessarily broad restart.
1. The 50-second route: locate, test, repair, recombine
Preserve the original attempt. Find the last point where the reasoning is clearly valid and the first point where it no longer supports the task. Then ask what could explain that transition. A visible sign error, for example, might be a transcription slip, an uncertain distribution rule or a consequence of an incorrectly formed expression.
Use a small fresh check to distinguish the plausible explanations. Do not immediately show the corrected step and count the learner’s agreement as diagnosis. Once the missing capability is identified, teach or practise it at the smallest useful scale.
Then put it back into a complete problem. A step that works in isolation may still fail when the learner has to select it, combine it with other operations or use it under time. The repair is not finished until it has been tested in the context where it matters.
The route is therefore not “always do less.” It is “make the size of the work match the size of the demonstrated problem.” A small repair is efficient when the surrounding capabilities are secure. A broader rebuild is appropriate when the evidence shows several connected gaps.
Do not let a chapter heading make that decision automatically. The heading tells you where the question belongs in a book. The original working and the follow-up evidence tell you what the learner needs next.
2. Why whole-topic restarts feel reassuring
Starting again from page one creates order. The student knows what to do, can tick boxes and often experiences early success because the opening examples are familiar. Parents can see a substantial amount of work being completed. The activity feels responsible.
That can be useful when the topic genuinely lacks structure. But it can also hide avoidance of the precise point where understanding is uncertain. The learner spends most of the session on secure material and reaches the difficult step only when tired or short of time.
There is another reason for overbroad revision: feedback often arrives at topic level. A score report may say “weak in trigonometry,” although the script shows accurate trigonometric modelling followed by unreliable algebra. The label describes the location of the question, not necessarily the source of the error.
Replace the broad instruction with a testable statement. “Revise trigonometry” becomes “check whether I can isolate the variable after selecting the correct trigonometric ratio.” The new statement does not assume the diagnosis is already certain. It identifies what the next small task should investigate.
The student still needs a broad syllabus map. This article is not an argument against organised coverage. It is an argument against using coverage as the automatic response to every local failure. Revision becomes more useful when broad planning and narrow repair are allowed to do different jobs.
3. The first wrong line is a location, not yet an explanation
Teachers often advise students to find the first mistake. That is a useful beginning, but the visible line may have several causes.
Suppose a learner writes −3(x − 4) = −3x − 12. The first visible error is the sign on the constant term. It might reflect an unstable rule for multiplying two negatives. It might be a copied sign despite intact knowledge. It might occur only when the learner rushes a long solution.
These causes suggest different repairs. A conceptual misconception needs explanation and contrasting examples. A transcription pattern needs an execution check. A time-related pattern needs attention to fluency and performance conditions as well as the operation itself.
Ask the learner to explain a fresh example before teaching the answer. Compare −3(x − 4) with −3(x + 4), using values that make the distinction visible. If they handle both correctly and can explain the signs, inspect whether the original error was isolated or recurrent in longer work.
Do not turn a single correct follow-up into proof that the original was “just careless.” The follow-up is one piece of evidence. Look at a small pattern across appropriate tasks.
The important shift is from naming an error to testing an explanation. “Wrong sign” describes the page. “The distribution rule is unstable when the second term is negative” describes a possible learning job that can be checked and repaired.
4. Preserve the original work before hindsight improves it
Once the model answer is visible, the learner can often reconstruct the route fluently. That is useful for learning, but it can obscure what was actually available during the first attempt.
Keep the original response. Add corrections separately. Record where help entered: a broad cue, a first-step hint, a worked example or the complete answer. This prevents supported reconstruction from being mistaken for independent diagnosis.
Kai Kai initially says he knew the correct transformation all along. His original line and a fresh parallel item show that the rule is still uncertain. The purpose of preserving the work is not to catch him being dishonest. Hindsight can make a corrected route feel obvious to anyone.
Alicia’s science answer presents the correct observation but no mechanism. After reading the scheme, she adds the mechanism immediately. That proves she can understand the correction in that moment. A fresh context is needed to see whether she can supply the relationship without the scheme.
The guide on self-marking without inflating the score develops this separation. For local repair, the same principle protects the diagnosis: keep the before-help evidence distinct from the after-help response. Otherwise every error starts to look smaller and easier than it was when the learner had to make the decision alone.
5. Separate the task into meaningful operations
A long question often contains a sequence: read the instruction, identify relevant information, represent the relationship, select a method, execute it, interpret the result and complete the response. Not every task uses that exact order, but the sequence helps locate different kinds of difficulty.
The operations should be meaningful, not microscopic. Splitting a sentence into every letter or a calculation into every pen movement usually adds little value. Split where the learner makes a decision or applies a relationship that could succeed or fail independently.
For a word problem, forming the equation is different from solving it. For an explanation, selecting evidence is different from stating the mechanism. For an essay paragraph, choosing an example is different from explaining why it supports the claim.
Once the parts are visible, ask which ones are already secure. A student who can model the problem accurately should not automatically spend most of the repair session relearning the model. A student who cannot model it should not be given only arithmetic drills because the last line contains a numerical error.
The map is a temporary diagnostic aid. The eventual performance must become integrated again. Students should not be left feeling that every question requires a long written process chart. The map helps choose the next teaching action; later practice should allow the learner to use the connected skill naturally.
6. A mathematics example: correct model, unstable transformation
Consider this original problem: three identical packages each contain two items costing x currency units each, with a discount of 5 currency units per package. The total paid is 33. A suitable equation is 3(2x − 5) = 33.
Kai Kai forms that equation correctly. He then writes 6x − 5 = 33 and solves from there. The modelling was successful. The distribution of the factor across the bracket was not. Restarting the entire word-problem chapter may spend too little time on the actual operation.
A diagnostic follow-up can remove the story and ask him to expand 4(2y − 3). If he writes 8y − 3, the pattern suggests the multiplier is being applied only to the first term. Explain why the entire bracket is multiplied, using a numerical check or a representation appropriate to the learner.
The corrected original route is 6x − 15 = 33, then 6x = 48 and x = 8. Substitution checks the model: each package costs 2(8) − 5 = 11, and three packages cost 33.
Now change the task. Use a different context in which the learner must again form and solve a bracketed equation. If the expansion works only after the teacher says “remember to multiply both terms,” the repair still depends on a prompt. The new capability must survive the full sequence, not merely an isolated expansion exercise.
7. A similar final error can begin with a wrong model
Now consider a different learner who writes 3(2x) − 5 = 33 for the same package problem. Their algebra may then be executed correctly. The failure happened before expansion: the discount was applied once to the total rather than once per package.
Giving this student distribution drills would miss the main issue. They need to distinguish a per-package reduction from a one-time reduction. A diagram, a repeated numerical example or a comparison of two short stories can make that relationship visible.
Ask: if one package costs 11 after its own discount, what should three packages cost? Then compare with a scenario where there is one discount on the whole purchase. The expressions differ because the situations differ, not because one is a preferred algebraic style.
This pair of learners shows why the final wrong answer cannot determine the repair. Both may lose credit on the same question. One formed the right model and transformed it incorrectly. The other formed the wrong model and transformed it correctly.
The topic name and final score are identical; the learning jobs are not. A useful revision session begins by identifying which relationship the student actually needs. That is the difference between doing more work in the right chapter and doing the work that changes the next attempt.
8. Check a prerequisite without turning it into a whole new course
A local failure can reveal an earlier prerequisite. If Kai Kai cannot distribute a multiplier reliably, the teacher may need to inspect multiplication, negative numbers or the meaning of brackets. That does not automatically mean every arithmetic topic must be restarted.
Trace the dependency only as far as the evidence requires. Use a simple numerical bracket, then a symbolic one, then a sign-sensitive version. Each step can help identify where the relationship becomes unstable.
If the numerical case is also unclear, conceptual teaching should begin there. If numerical reasoning is secure but notation causes confusion, focus on representation. If both work in isolation but fail inside long solutions, practise integration and checking under appropriate conditions.
A prerequisite check should therefore be a branching investigation rather than a predetermined march backwards through the textbook. It asks which earlier capability is missing, not how far back the student can be sent.
This protects both time and confidence. The learner can see that some capabilities are intact while another needs work. It also prevents false reassurance: if the check reveals a genuinely broad foundational gap, the teacher has evidence to justify a larger rebuild.
The aim is neither to minimise the problem cosmetically nor to maximise the amount of teaching. It is to find the smallest coherent foundation from which the learner can successfully reconstruct the target skill.
9. Use questions that separate plausible explanations
A good diagnostic question is chosen for the information it can produce. If two explanations would lead to the same response on the follow-up item, that item may not help decide between them.
For example, a learner who confuses area with perimeter might still answer a specially chosen numerical case correctly by coincidence. Choose dimensions where the quantities clearly differ and ask for an explanation of what each measures. The goal is to expose the distinction, not merely obtain another right or wrong number.
EEF’s guidance on diagnostic assessment in mathematics discusses approaches such as hinge questions, questioning and low-stakes checks. It emphasises deciding why the assessment is being used and how the information will guide subsequent teaching. That practical orientation is important here.
Before giving the follow-up, state privately what different responses would suggest. If the learner handles the concept but miscopies a value, investigate execution. If they choose an inappropriate relationship consistently, teach the concept. If they succeed only after a cue, inspect access and independence.
Do not overclaim diagnostic certainty. A small question is a sample. It can narrow the possibilities and justify a next teaching step without proving a complete theory of the learner. The best probe is useful enough to guide action and limited enough not to consume the time needed for the repair itself.
10. Do not teach the answer while claiming to test the gap
A teacher asks, “You need to multiply the second term too, don’t you?” The learner agrees and corrects the line. That may be helpful feedback. It is not an independent check of whether the learner knew to distribute the multiplier.
Separate diagnosis from instruction. First ask a fresh question that gives the learner a fair opportunity to show the relationship. Then teach if needed. Afterwards, use another task to check whether the teaching has changed performance.
This sequence need not be long. A few carefully chosen questions can provide enough evidence for a small intervention. The aim is not to withhold help for the sake of purity. Once the gap is clear, help should be specific and prompt enough to support learning.
Record the support condition honestly. “Correct after a first-step cue” is different from “correct independently.” Neither is a moral judgement. The difference tells the teacher what still needs to be returned to the learner.
Parents can use the same distinction at home. Instead of supplying half the answer and then saying the child knew it, ask what they can do before help. Then provide appropriate support without pretending that the supported success proves independent readiness. A good repair process can contain both generous teaching and honest measurement.
11. Science: facts can be secure while the causal bridge is missing
Alicia knows that warmer water can change how quickly a substance dissolves under specified conditions. In an explanation task, she repeats the observation but does not state the relevant mechanism expected in her course. Rereading every fact in the topic may not address the missing explanatory relationship.
First check the concept at the appropriate curriculum level. Can she explain the process orally without the answer being supplied? Can she distinguish an observation from an explanation? Does she know which variables were held constant in the described comparison?
If the mechanism is unknown, teach it accurately. If she can explain it but omits it in writing, practise converting the relationship into a concise written answer. If she selects the wrong evidence, repair evidence selection before demanding a better conclusion.
Use a fresh context after the focused work. The student should not merely memorise one sentence attached to one diagram. Ask for an explanation in a changed but syllabus-appropriate situation and inspect whether the causal bridge appears independently.
The same visible mark loss can therefore come from missing knowledge, weak selection or incomplete expression. The repair should follow that distinction. A student who needs one explanatory connection should not be required to copy an entire chapter before being allowed to practise it. A student who lacks the underlying mechanism should not receive only a sentence frame and be declared fixed.
12. Reading: the weak step may be reference, not inference as a whole
Consider this original passage: “Maya handed the folder to Lina. She placed it beside the blue bag, then returned to her desk.” Without further context, the reference of “she” may be ambiguous. A good reading task must provide enough information for the intended answer rather than asking the learner to guess what the writer meant.
Now revise it: “Maya handed the folder to Lina, who placed it beside the blue bag before returning to her desk.” The relative clause identifies Lina as the person performing the action. A learner who misreads that relationship may later make an incorrect inference about Maya’s behaviour.
The final inference can look weak even though the first failure was reference tracking. Before assigning a large set of inference questions, check whether the learner can identify who did what in the relevant sentences.
If reference is secure, inspect the next step: selecting evidence and connecting it to a justified conclusion. If that relationship is weak, practise it directly. The repair should move along the actual reading chain rather than treating every wrong comprehension response as the same broad weakness.
This example also reminds teachers to inspect their own materials. An ambiguous practice sentence can create a false diagnosis. The learner should not be sent to relearn grammar because the question itself does not support a unique answer. Diagnostic quality depends on the clarity of the task as well as the student’s response.
13. Writing: repair the argument link without replacing the writer
Tricia’s paragraph contains a clear claim and a relevant example. The feedback says the example is not explained. Her response is to discard the paragraph and copy a model version. The new paragraph reads better, but it may not teach her how to connect her own evidence to her own claim.
Keep the useful parts visible. Ask what the example shows and why that matters to the claim. A missing bridge might be one or two sentences of reasoning, not a completely new argument.
For an original example, a student claims that a school should make its library opening times more predictable. They cite pupils arriving after lessons to find the library unexpectedly closed. The explanation should connect unpredictability to planning and access, rather than merely repeat that the library was closed.
If the claim itself is unclear, or the example irrelevant, the repair needs to expand. Do not force a narrow link exercise onto a paragraph whose structure does not support it. The teacher should preserve what works while honestly identifying what does not.
After the focused revision, use a new prompt. Can Tricia choose evidence and explain its relevance without receiving the connecting sentence? That is the transfer question. A polished corrected paragraph is useful, but the capability belongs to the learner only when they can perform the relationship in fresh writing.
14. Worked examples can isolate a step, but support must fade
A partially worked example can make a specific operation visible. The teacher may supply secure earlier steps and leave the uncertain transformation for the learner. This reduces the amount of unrelated work required while the target is being taught.
Research on worked-example fading is relevant, but it should not be overstated. Miller-Cotto and Medrano studied several instructional conditions with sixth-grade geometry learners. Their findings support examining how worked support and fading are designed, while also showing that prior knowledge matters. The study does not validate every possible one-step repair routine.
Use the approach thoughtfully. If the student cannot understand the supplied earlier steps, leaving only the final operation blank may conceal a broader gap. If they can understand and reproduce those steps, a focused completion task may be a useful temporary scaffold.
Then remove the support gradually. Ask the learner to generate the previously supplied steps, choose the operation and complete a fresh problem. The endpoint is not repeated success on a worksheet that always points directly at the required move.
Support is successful when it helps the learner take over more of the task. It becomes misleading when the teacher counts supported completion as proof that the whole independent skill is secure. Keep the distinction clear in both practice and reporting.
15. Build a repair set that tests the rule, not one answer
A useful local repair set includes examples, contrasts and at least one changed form. For distribution, use positive and negative multipliers, different term orders and a numerical check. For reference tracking, vary sentence structure while keeping the referent clear. For explanatory writing, use different evidence supporting different claims.
Do not add variation randomly. Each change should test whether the learner understands the relationship that previously failed. If every feature changes at once, a new error may be difficult to interpret.
Begin with a clear example when the rule is not yet understood. Then let the learner attempt similar cases without prompts. Once those are stable, introduce the relevant contrast. The sequence can be adjusted to the learner’s prior knowledge rather than imposed mechanically.
Include a non-example where the tempting move would be wrong. A learner who has just practised distributing a multiplier should still recognise when no multiplier is outside the bracket. A learner practising causal explanation should still answer a description question without adding an unsupported cause.
The repair should therefore improve discrimination as well as execution. The student must know what to do and when to do it. A narrow drill that creates a new overgeneralised habit has solved one problem by creating another. The target is a reliable relationship that can be selected appropriately in fresh work.
16. Recombine before declaring the repair complete
A student can expand brackets correctly in a focused set and still fail to do so inside a longer word problem. In isolation, the worksheet announces the operation. In the complete task, the learner must recognise where it belongs while carrying other decisions.
Return to a fresh integrated problem. Do not simply repeat the corrected original until its route is memorised. Change the context or representation enough that the learner must reconstruct the solution while preserving the target skill.
For Kai Kai, the new task should require forming the model, expanding, solving, interpreting and checking. If expansion succeeds but modelling fails, a different weakness has become visible. Do not conclude that the expansion repair was useless; identify the next job.
For Alicia, a new science explanation should require choosing evidence and supplying the mechanism without a highlighted blank. For Tricia, a new paragraph should require her own claim and example before the explanatory bridge.
Recombination protects against an important illusion: making a step easy by removing everything around it can improve practice performance without proving whole-task readiness. The isolated task is a teaching tool. The integrated task checks whether the new capability can coexist with the rest of the work.
This is also where the size of the repair may need to expand. If several neighbouring steps fail repeatedly, a broader teaching sequence may be more coherent than a collection of disconnected micro-drills.
17. Delay the check and preserve some freshness
Immediate success after explanation is encouraging, but the solution may still be active from the teaching event. A later check asks whether the learner can retrieve and use the relationship after that immediate support has faded.
Use a fresh item after an appropriate gap. The gap should fit the schedule and learning goal; there is no universal interval for every skill. The important point is to include evidence beyond the moment immediately after correction.
Butler’s research on repeated testing and transfer provides relevant experimental evidence that retrieval practice can support transfer beyond repeated study in the tasks examined. It does not establish that one local drill guarantees examination improvement. The practical lesson is to check what the learner can produce later, not only what they recognise now.
Preserve some questions for independent checking rather than teaching every example in advance. The guide on saving unseen questions develops that resource decision.
Do not hoard all fresh material until the final week. Use it when the result can still change teaching. A timely small check can reveal that a correction remains prompt-dependent while there is room to repair it. Freshness is valuable because it makes the evidence more informative, not because novelty is an educational virtue by itself.
18. Timing can expose a different weak step
A learner may perform the repaired operation correctly when untimed but make errors inside a timed paper. That does not automatically mean the concept was never learned. It may indicate insufficient fluency, poor pacing or loss of a checking habit under pressure.
Add timing gradually after accuracy is reasonably stable. A short timed set can inspect execution cost. A mixed section adds selection. A full paper adds endurance and competing demands. Each stage answers a different question.
Do not time fragile knowledge so aggressively that the learner rehearses errors. When the rule remains unclear, return to teaching. When the rule is clear but slow, practise efficient retrieval and execution. When it is fluent alone but fails late in the paper, inspect integration and pacing.
The first visible error may move as learning improves. Previously, the student could not choose a method. Now they choose it but execute slowly. Later, execution becomes fluent and the final-answer interpretation becomes the remaining issue. The repair plan should move with that evidence.
This is why “topic fixed” can be too simple a label. A topic contains several capabilities operating under different conditions. The useful question is which part is now reliable and which part still limits the intended performance. Local repair should make that picture clearer rather than pretending one successful drill settles every future demand.
19. When a whole-topic rebuild is the right decision
Sometimes the title’s narrow scenario does not fit. A student may lack the central model, several prerequisites and the ability to recognise standard examples. In that situation, repeated one-step drills can become a fragmented substitute for understanding the topic.
Broaden the work when fresh checks show connected gaps across the structure. If a learner cannot explain the meaning of an equation, choose operations or interpret a solution, the repair may need a coherent lesson sequence rather than one correction at a time.
Broaden it also when the learner’s knowledge consists of isolated memorised procedures with no reliable way to select among them. The missing capability may be the organisation of the topic itself.
The decision should not be framed as failure of surgical teaching. Good diagnosis sometimes justifies major teaching. The principle is proportionality: small when the evidence supports small, larger when the evidence supports larger.
Explain the reason to the student. “We are rebuilding this section because the same relationship is unclear in several different tasks” is more informative than “you need to start again.” Preserve any secure capabilities within the rebuild so that the learner does not experience the process as erasing everything they have already achieved.
A coherent rebuild can still be selective. It need not repeat every page in the original order if some parts are already established. The topic structure should guide the teaching, while the learner’s evidence guides where time is spent.
20. Keep other learning alive while repairing the bottleneck
A focused repair can absorb attention so completely that previously secure skills disappear from the plan. Avoid turning one weak step into the only thing the learner practises for weeks.
Retain small independent returns to established capabilities, especially those needed inside the repaired task. If algebra is being repaired for calculus, occasional complete calculus questions can later check reintegration. If evidence-to-claim links are being repaired in writing, continue some ordinary reading and vocabulary work rather than replacing all language learning with one paragraph exercise.
The maintenance dose should remain smaller than the active repair when the evidence justifies that balance. Do not use comfortable strong-topic work to avoid the difficult step. Equally, do not delete all maintenance because the repair is urgent.
Schoolwork may already provide useful returns. Count genuine independent use rather than adding duplicate worksheets automatically. The plan should coordinate learning opportunities, not simply accumulate them.
When the repaired step becomes secure across fresh work, reduce the intensive block. Return it to ordinary practice and occasional checking. A repair programme should have an exit condition. Otherwise the student keeps paying the full cost of an intervention long after its original purpose has been achieved.
The goal is better whole performance, not permanent attention to one defect. The local repair earns its place by freeing the learner to use the rest of their knowledge more effectively.
21. Make the next revision instruction specific enough to act on
A useful instruction names the task, the target and the check. “Practise distribution with negative terms, explain the sign changes, then solve one fresh word problem without a prompt” is actionable. “Do more algebra” is not.
For science: “Distinguish the observation from the mechanism in two fresh contexts, then write one complete explanation.” For reading: “Track who performs each action in a clear passage, then answer a new inference question.” For writing: “Explain how the evidence supports the claim, then produce that relationship in a new paragraph.”
These instructions should not become scripts the student cannot leave. They are temporary plans for a demonstrated need. Once the capability is reliable, ordinary tasks should increasingly carry it.
A parent can ask the learner to explain the repair in one sentence. If the answer is only a topic name, the task may still be too broad. If the answer is a tiny action disconnected from any purpose, the repair may be too narrow. The learner should know both what is being practised and where it will be used.
This is an important form of independence. Students become better at selecting revision when they can distinguish a chapter label from a failed relationship, and a failed relationship from the whole of their ability. The next session then begins with a useful question rather than a vague instruction to start everything again.
22. A small record is enough to preserve the reasoning
Keep the original error, the tested explanation, the repair task and the fresh verification result. These four elements preserve why the plan changed without creating a museum of every wrong answer.
For Kai Kai, the record might identify distribution across both terms as the target, note that modelling was independently correct, and record whether the operation survived a fresh complete problem. For Alicia, it might distinguish missing mechanism knowledge from missing written expression. For Tricia, it might identify the evidence-to-claim bridge while preserving the useful argument already present.
Do not store more personal information than the teaching decision needs. There is no need to label the student with a permanent weakness or turn a short error pattern into a public narrative. The fictional cases in this article illustrate mechanisms; real learners deserve privacy and room to change.
Review the record when the next similar error appears. It may be the old cause returning, a different cause producing a similar symptom or an isolated slip. Do not assume that a matching red mark proves a matching diagnosis.
The record should support continuity, not rigidity. Its value is that a teacher or learner can remember what was actually tested and what remained uncertain. That is more reliable than a vague recollection that the topic was once “done” or that the child is “always careless.”
23. The smallest useful repair still serves the whole learner
A question belongs to a topic, but an error belongs to a particular chain of decisions. Sometimes the whole chain needs rebuilding. Sometimes one relationship is missing while much of the surrounding capability remains intact.
Preserve the original work. Locate the break. Test a plausible explanation without quietly supplying the answer. Teach the missing relationship. Practise it with enough variation to reveal understanding. Then recombine, delay and check under the conditions where it matters.
Do not use the promise of efficiency to minimise genuine difficulty. A student with broad gaps deserves substantial teaching. Do not use the seriousness of education to justify unnecessary repetition either. A learner who needs one transformation repaired should not have to repeat a chapter merely to make the work look substantial.
The practical question is: “What is the smallest coherent change that could improve the next complete attempt, and what evidence would show it worked?” That question keeps both precision and uncertainty in view.
Revision should leave the student more capable, not merely more familiar with the opening pages of the textbook. When the repair fits the failure, the next session can become smaller, clearer and more useful without lowering the standard of the final performance.
Sources and further reading
EEF: Five Ways to Use Diagnostic Assessment in the Mathematics Classroom explains the connection between diagnostic questions and subsequent teaching decisions. Miller-Cotto and Medrano: Does Working Memory Moderate the Effect of Fading on Math Performance? examines worked-example support in sixth-grade geometry. Butler: Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying provides experimental retrieval-and-transfer evidence. None tests this article’s entire proposed repair routine.
Continue through How Learning Diagnosis Works, the Learning Runtime Hub, or the Complete Examination Craft Index, according to whether the next job is diagnosis, teaching or examination deployment.
