How Bukit Timah Tuition Does Not Work (Bukit Timah Tuition OS / Education OS, failure-first) — v1.1

Bukit Timah tuition does not work for a particular student when the lessons, practice and feedback do not address that student’s actual learning need at a sustainable cost. A child may attend conscientiously, finish every worksheet and still struggle to begin an unfamiliar question. Another may improve substantially while continuing to need specialist teaching. These are different situations. Before deciding that tuition is ineffective, distinguish the learning problem, the support being provided and the evidence of change.

Parents searching for why tuition is not helping their child, why grades are not improving despite tuition, or when to change a tutor need more than a list of warning signs. They need a fair way to examine reading, vocabulary, Mathematics, Science, homework independence, examination timing and the whole school week. This guide explains how a mismatch develops, what to look at in real work, and how to choose between continuing, adjusting, reducing or replacing support.

Bukit Timah is the context, not the explanation for every difficulty. This is not a ranking of local centres or an allegation about named providers. It is a parent-facing guide to possible teaching and learning failures, using original examples and the eduKate learning library. The central rule is to judge progress on a clearly identified capability under fair, comparable conditions—not by prestige, worksheet thickness, one test result or a compulsory reduction in all help.

Reviewed against the sources linked in this guide on 20 September 2026. Alicia, Tricia and Kai Kai are fictional teaching characters; their conversations, marks and timetables below are illustrative, not testimonials or research data.

Your 50-second route through the problem

Choose one piece of recently taught work. Ask what the child was expected to learn, what help was available and what they can now do on a fresh but comparable example. Keep necessary accessibility support. Look for the first specific difficulty: understanding, remembering, choosing, executing, explaining, checking or working within an appropriate time. Then ask the tutor for one change that targets that difficulty and a reasonable date to review new evidence. Do not remove all support or add another class simply to see what happens. A useful review should make the next decision clearer and the child safer to teach, not produce another label.

Open the chapter guide

1. Define success fairly · 2. What marks can and cannot show · 3. Build a comparable baseline · 4. Explanation versus generation · 5. Find the actual prerequisite · 6. Reduce unneeded prompting · 7. Method selection and transfer · 8. English problems that look alike · 9. Science reasoning and evidence · 10. Examination timing · 11. Homework and correction · 12. Lesson and group fit · 13. The weekly cost · 14. Parent boundaries · 15. School alignment · 16. Fair access and necessary support · 17. Digital tools and AI · 18. Read progress without overclaiming · 19. Continue, adjust, reduce or change · 20. Speak with the provider · 21. Examination-season decisions · 22. A practical review plan · 23. Questions parents ask · 24. Sources and further routes.

1. Define success before declaring that tuition has failed

The sentence “tuition is not working” can contain several different complaints. A parent may mean that the latest mark is unchanged. The child may mean that the explanations are difficult to follow. The tutor may believe a foundational gap has improved, while the school assessment has moved to a different topic. Another family may be satisfied academically but unable to sustain the travel and homework. None of these interpretations should be silently substituted for the others. Begin by naming the outcome that the arrangement was intended to produce.

For one student, the immediate outcome is access: being able to understand the questions and participate in school lessons. For another, it is reliable calculation, more coherent writing or a better way to revise. An advanced learner may seek challenging discussion rather than a higher mark on already-secure school content. A student returning after illness may need temporary catch-up. The same lesson format cannot be judged sensibly without knowing which of these jobs it is supposed to perform.

Write the purpose as an observable sentence. “Improve Mathematics” is too broad to guide next Tuesday’s lesson. “Form an equation from a verbal relationship without being given the operation” is usable. “Become confident in English” needs interpretation. “Choose a relevant textual detail and explain the inference it supports on a fresh passage” gives teacher and learner something to examine. The purpose can change as learning develops, but the current purpose should be visible.

Do not make declining tuition hours the sole definition of success. A student can become more independent with linear equations while beginning a new topic that needs substantial teaching. A child can learn to use assistive technology more effectively while continuing to require it. Another may choose a long-running enrichment programme because the activity remains valuable. Continuing support is not, by itself, evidence that teaching has failed. Ask whether the support remains appropriate and whether the particular capability being taught is developing.

The research picture also does not justify treating tutoring as generally ineffective. A systematic review by Andre Nickow, Philip Oreopoulos and Vincent Quan found positive average effects across experimental tutoring studies. That supports taking well-designed tutoring seriously; it does not establish that every programme, delivery model or local provider will help every child. A research average is not a guarantee for an individual family. Read the authors’ research summary.

A sensible definition therefore has three parts: a worthwhile learning goal, evidence that the learner is moving towards it, and a cost the household can reasonably sustain. Cost includes time and effort, not only fees. Progress may be modest but important. A formerly inaccessible question may now be attempted. A student may still make mistakes but identify the relevant method independently. The review should recognise those changes while remaining honest about what has not improved.

Imagine Alicia has attended four lessons focused on understanding fraction size. Her next school paper includes geometry that was not part of those lessons. A flat overall score cannot settle whether the fraction teaching helped. Conversely, a higher overall score cannot prove that it helped if the paper happened to contain easier questions elsewhere. Examine the fraction capability directly, then return it to broader school work. Keep the narrow evidence and the broad outcome connected without pretending they are identical.

This is a review of an arrangement, not a trial of a child’s worth. Avoid language that makes the learner responsible for proving that adults spent money wisely. The child should be able to show confusion without fearing that every error will trigger disappointment, another test or another class. An honest account of difficulty is useful information. A polished worksheet completed through hidden help is much less useful, even when it looks reassuring.

2. What a mark reveals—and what it leaves out

A mark summarises performance on a particular collection of tasks under particular conditions. It may combine recall, reading, reasoning, calculation, written expression and time management. The total is important when it contributes to a formal result, but it is a compressed description. To decide what to teach next, unpack at least some of the work underneath it. A score cannot tell you automatically which process caused the loss.

Suppose two students each receive 60 out of 100 on an illustrative Mathematics paper. One attempted every question but misunderstood percentage change. The other solved attempted questions accurately but left a substantial section blank. The totals match; the likely teaching priorities do not. Giving both students the same thick set of percentage worksheets would address only one possible problem. Giving both speed drills would be equally poorly matched.

Now compare one student’s 60 in January with 65 in March. Before calling the five-point difference a tutoring effect, ask whether the content, difficulty, marking and testing conditions were comparable. There may have been genuine improvement. There may also have been changes in the paper, school teaching, attendance, practice or chance variation. A family review need not become a formal research study, but it should resist assigning the whole change to the most visible intervention.

Separate the overall score from a few useful process observations. Did the learner choose a method before asking for help? Was the answer responsive to the question? Did the student recover after a wrong start? Could they explain the correction later? Were the questions that had previously been inaccessible now attempted sensibly? These observations do not replace marks. They help explain them and make the next instructional choice more precise.

Be careful with percentages from tiny samples. A change from two correct answers out of three to three out of three looks like a large percentage increase, but it is one additional item. Record the actual counts, the questions and the support conditions. A short probe is useful for deciding what to inspect next, not for declaring a durable transformation. Repeated evidence across suitable tasks is more informative than a dramatic percentage from one small set.

A high mark can coexist with narrow preparation. That possibility is worth checking, but it is not a reason to dismiss all high marks as superficial. Examination preparation often legitimately includes learning common forms, notation and response expectations. The question is whether the learner understands enough to recognise when those forms apply and when they do not. Familiarity is useful; dependence on one exact surface is the potential limitation.

A lower mark can coexist with real learning too. A student may be attempting more demanding work, moving to a different subject level or completing the task with less instructional help. These changes must be recorded. They do not make the lower result irrelevant, but they change its interpretation. Never compare a heavily prompted easy task with an unprompted harder task and conclude that independence training has made the child worse.

For a practical paper review, select three losses rather than dissecting everything at once. Choose one repeated error, one question the learner could not start and one time-related problem. Ask what each loss suggests and what further evidence would distinguish the possibilities. This produces a small teaching agenda. The remaining paper can stay as context instead of becoming a prolonged family post-mortem.

The useful end point is a sentence such as: “The overall mark is unchanged, but the student now sets up direct equations independently; translating comparisons remains weak.” That statement gives credit where evidence supports it, preserves the unresolved issue and identifies a next step. It is more informative than either “tuition works” or “tuition does not work” on its own.

3. Build a fair comparison before making a large decision

A review needs a starting point, but the starting point should be modest. Choose a capability that has actually been taught and matters to current learning. Record what the student can do with the support normally appropriate to the task. Then select a fresh example of similar demand. Do not use an unseen advanced question simply because it looks rigorous. A task outside the learner’s taught knowledge cannot fairly test whether recent teaching has been retained.

State what will stay the same. For a calculation check, keep the operation and approximate arithmetic difficulty similar. For reading, keep passage length and language demand reasonably comparable. For writing, use prompts with similar planning demands rather than comparing a familiar personal narrative with a difficult argumentative question. Exact equality is rarely possible in ordinary practice, so note the differences instead of pretending they do not exist.

State what will change. Perhaps the values are different, the story is new or the question no longer names the method. Change only enough to examine the capability you are interested in. When wording, topic, time, support and difficulty all change at once, a weaker result becomes difficult to interpret. The child may have learned the original skill and still be encountering several new demands.

Keep access conditions appropriate. Glasses, required assistive tools, agreed breaks and approved assessment arrangements are not interchangeable with a tutor giving the next algebraic step. A fair review distinguishes assistance that enables access from assistance that supplies the answer or method being assessed. Coordinate uncertainties with the school or relevant professional rather than improvising a removal test.

Use a brief record: task, date, taught content, help provided, outcome and next question. “Needed help” is not precise enough. “Asked the adult to explain the word remainder; then chose division independently” is different from “adult said to divide by six.” The first may concern language access; the second supplies the mathematical choice. A few accurate words are better than a complicated colour-coded score.

Choose a review interval that fits the learning and the school calendar. A later return can show whether knowledge remains accessible, but there is no universal number of days that proves mastery. A review immediately after teaching mostly shows what is available then. A review much later may include new learning, illness or a curriculum shift. Both can be useful when their limits are understood.

Do not secretly examine the student. Tell them what the short check is for: “We are checking whether this explanation helped you choose the operation. This is not another grade.” Allow an honest attempt and stop when the task has revealed the relevant difficulty. The aim is to improve teaching, not to see how long the child can tolerate being stuck. Consent and cooperation are especially important when a family has already had repeated homework conflict.

Keep one comparison unchanged when the rest of the programme advances. For example, retain a small direct-equation check while introducing word problems. This lets the tutor see whether the established skill remains stable as new demand is added. Without an anchor, every lesson may be harder than the last, making progress look absent because the learner is always being judged at the edge of current capability.

The separate Tuition Inversion Test guide develops a fuller observation procedure. Here the important principle is comparability. A fair comparison protects students from false failure claims and protects families from false success claims. It also makes good teaching easier to recognise.

4. When an excellent explanation does not become usable learning

A clear explanation can be valuable without being sufficient. During a demonstration, the adult chooses the starting point, names the relationship, decides the order and notices mistakes. The student may understand each step as it appears. Working alone asks for something additional: retrieving the relevant knowledge and deciding which step to generate. When the lesson never creates that transition, both tutor and student can overestimate what is ready for independent use.

Consider the equation 3x + 5 = 20. The tutor says, “Subtract five from both sides, then divide by three.” The student follows and obtains x = 5. That is a useful supported success. It does not yet show whether the student recognises the inverse-operation structure. A fresh equation, 4y + 3 = 19, offers a small opportunity to inspect the decision. Ask the learner to identify what happened to the unknown and how to undo it.

The purpose is not to ban demonstrations. An unsupported learner may need a full model. The useful sequence is to make the reasoning visible, let the student complete a related step, and then create an independent attempt with appropriate feedback. The amount of support should follow the learner’s response rather than a rigid rule that every lesson must contain a prescribed number of minutes of silence.

The What Works Clearinghouse study guide recommends alternating worked solutions with problem-solving attempts and connecting concrete and abstract representations. It also distinguishes recommendations by evidence strength. That is a more careful foundation than claiming that all direct explanation is passive or that all difficulty is productive. See the guide’s recommendations and their evidence ratings.

A useful tutor can ask a student to predict one step before revealing it. “What should happen to both sides now?” is a different demand from “Copy the next line.” The student may answer incorrectly; that is information the model alone would have hidden. The tutor can then explain the decision, use another example and check whether the learner can make it. This creates an exchange between instruction and evidence rather than a lecture followed by a vague question about understanding.

For writing, the same distinction appears when the tutor builds the paragraph aloud. The child can agree that a sentence fits while being unable to decide what their own next sentence should accomplish. Instead of asking for an entire essay immediately, ask for one paragraph purpose: describe a change, explain a reason, support a claim or show a consequence. Then ask the student to generate a sentence serving that purpose. The instructional target becomes visible.

For vocabulary, recognising a definition in a list differs from retrieving a word for an appropriate sentence. Suppose the student recognises cautious as careful to avoid risk. Ask them to explain the difference between cautious and frightened in a short situation. Then ask for a sentence in which cautious fits but frightened would change the meaning. The task tests usable meaning rather than whether the student remembers seeing the definition.

A parent can review this without observing the whole lesson. Ask the tutor what the learner produced independently near the end and what will be revisited later. Ask the child which decision they can now make that previously required help. Neither answer needs to be impressive. A precise small gain is more useful than an elaborate report that does not identify any student-generated work.

When a child says, “I understand with the teacher but not alone,” treat that as a starting description, not proof of manipulation or failure. The next teaching task may be retrieval, method selection or simply another supported step before release. Good review does not shame the student for accurately reporting the boundary of current learning. It uses that boundary to decide what instruction should do next.

5. Repair the prerequisite that is actually blocking the task

A new chapter can expose an old difficulty. The visible topic may be percentage change, but the first problem may be identifying the reference quantity. A calculus answer may fail through fraction manipulation rather than the differentiation rule. A comprehension response may fail because a key word in the question is unknown. Describing every loss by the chapter title can lead to broad reteaching that misses the shared cause.

Start with the earliest step the student cannot justify. Do not automatically work backwards through an entire syllabus. A short prerequisite check can distinguish possibilities. For a percentage problem, ask what 20% of 50 means, then ask what quantity a stated percentage refers to. A student who calculates the first accurately but misidentifies the second needs a different explanation from a student who cannot connect percent to one hundred at all.

Worked example: an increase and a decrease are not interchangeable

In this original example, a price rises from $80 to $100. The increase is $20. Relative to the original $80, the increase is 20 ÷ 80 = 0.25, or 25%. Now reverse the journey: a price falls from $100 to $80. The decrease is still $20, but the reference quantity is now $100. The decrease is 20%. The arithmetic difference is simple; the conceptual difference is the base.

A student who says both changes are 20% may be dividing every change by the larger number. Another who says both are 25% may be dividing by the smaller number. Neither rule is reliable. Ask, “Percentage of what?” before calculation. Mark the original amount in each sentence. Then write change divided by original amount. The repair is not memorising two answers. It is learning to name the reference quantity before choosing the denominator.

Use a fresh pair: a quantity rises from 40 to 50, then falls from 50 to 40. The answers are again 25% and 20%, but the student should be able to explain why. Next use a pair with different ratios, such as 60 to 90 and 90 to 60. The increase is 50%; the decrease is one third of the original, or 33⅓%. This checks the decision without relying on the first numerical pattern.

Re-enter the original school problem after the small repair. A prerequisite lesson is not complete merely because the student succeeds on isolated examples. Ask them to identify the reference amount in a realistic question containing extra information. If that works, the intervention has restored access to the current task. If it does not, inspect whether reading or another relationship is now the first difficulty.

Worked example: algebra inside a more advanced method

Suppose the learner correctly obtains an intermediate expression, 2(x + 3) − 3(x − 1), but simplifies it to −x + 3. The advanced topic that produced the expression may not need reteaching. Expanding carefully gives 2x + 6 − 3x + 3 = −x + 9. The error is in applying the negative multiplier to the second bracket. A short contrast with 2(x + 3) + 3(x − 1) can make the sign decision explicit.

Check the simplification by substitution. At x = 0, the original expression is 6 − 3(−1) = 9. The incorrect expression gives 3. This does not prove every symbolic identity from one value, but it disproves this claimed identity immediately. Teach the student what the check can establish. A counterexample can refute an incorrect simplification; one matching value cannot guarantee a general identity.

The value of a narrow repair is that it can improve several later tasks without creating a separate programme for each one. However, be cautious about claiming a single root cause explains everything. Students can have several independent difficulties. The first repaired prerequisite may reveal another. A working hypothesis should become more precise as evidence arrives, not turn into a theory the tutor defends regardless of the student’s responses.

For families, the practical question is: “Which earlier skill is blocking the current work, how are we checking that, and when will the child return to the current task?” That question protects against both extremes—rushing ahead without foundations and keeping a learner indefinitely in simplified exercises that no longer connect to school. The existing learning-repair manual provides a broader route for matching repair to the observed problem.

6. When helpful prompting becomes unnecessary prompting

Prompting is not inherently bad. It can help a learner notice a relevant feature, recall a strategy or complete a step that is not yet independent. The question is whether the prompt supplies the decision currently being assessed and whether the teaching plan gives the student an appropriate opportunity to make that decision later. Calling all assistance dependence confuses teaching with a test of unsupported performance.

In an illustrative lesson, Alicia sees a word problem and waits. The tutor immediately says, “Draw a bar model.” Alicia draws it correctly. A parent watching the final page sees successful work. The missing information is whether Alicia could have chosen the representation herself. The next lesson should not begin with an accusation. It should create a small opportunity: ask Alicia to state the quantities and suggest a representation before a method is named.

Prompts can be recorded by function rather than by a supposedly scientific numerical scale. A neutral instruction might repeat the task. A general strategy cue might ask the student to identify what is known. A more specific clue might name a relationship. A model shows part or all of the solution. These levels are not equally informative about independence. Record the actual wording when it matters; different adults can interpret the same label differently.

Wait long enough for a genuine attempt, but do not turn waiting into a test of endurance. Silence may mean the learner is thinking, cannot read the task, is unsure how to ask for clarification or does not know the method. The tutor can ask a neutral process question to distinguish these possibilities. The aim is to preserve useful thinking while avoiding prolonged, unproductive frustration.

EEF’s metacognition guidance describes explicit teaching and scaffolding of planning, monitoring and evaluation, with increasing independent use. This supports a graduated approach rather than abrupt withdrawal. The important distinction is between teaching a strategy and assuming the child will acquire it merely by being left alone. Read the second-edition guidance.

Use a small release sequence on comparable tasks. First the tutor models why a representation fits. Then the student chooses between two possible representations and explains the choice. Next the student proposes a representation without a menu. Finally, the learner uses the strategy in a mixed set. A step backwards may be appropriate when the task changes substantially. The sequence is a teaching option, not a mandatory staircase every child must climb at the same speed.

Parents sometimes add prompts without noticing. Reading the question with emphasis, pointing to a number, raising an eyebrow at a wrong answer or asking “Are you sure?” only when the child is wrong can all supply information. A home check intended to observe independent work should make these signals explicit. During teaching, they may be useful. During a comparison, they change what the result means.

Do not punish help-seeking. A student who says, “I do not understand what consecutive means here,” is identifying a barrier precisely. That is different from asking the adult to do the question. Good independence includes knowing when clarification is needed and using available support appropriately. The goal is not a child who never asks a question; it is a child who increasingly understands the question they need to ask.

After a few suitable opportunities, review whether the same task decision requires less instructional help. Compare like with like. A student may need fewer prompts on familiar equations and more on a new representation. That can be a healthy learning pattern. The meaningful claim is narrow: “She now selects the operation on this taught problem family without a method cue.” Avoid turning it into “She is independent in Mathematics” before the evidence supports that broader conclusion.

7. Why repeated worksheets can miss method selection and transfer

A chapter heading can do part of the student’s thinking. On a worksheet labelled simultaneous equations, the learner does not need to decide whether simultaneous equations are relevant. On a school paper, that label may be absent. The student must recognise the structure before executing the method. Strong chapter performance and weak mixed performance can therefore coexist without contradiction.

Begin by separating selection from execution. Give a small set of questions and ask only which method or relationship seems appropriate, with a short reason. Do not require full solutions immediately. This produces several meaningful decisions without the time cost of completing every calculation. If selection is weak but procedures are strong, the next practice should target discrimination rather than simply increase the number of same-topic exercises.

Use contrasts with a purpose. “Find the area of a rectangle with length 8 and width 5” and “A rectangle has area 40 and width 5; find its length” share the same relationship but ask for different unknowns. A student who multiplies in both cases may recognise the topic without representing the target. Ask what is known, what is unknown and how the quantities are related before naming an operation.

Transfer adds a related challenge: recognising a familiar relationship when the surface changes. An area relationship can appear through tiles, a garden, a plan or an algebraic expression. A comparison can be written verbally, shown in a table or embedded in a larger problem. The goal is not to expose the child to every possible story. It is to help them identify the features that matter and separate those from decoration.

Do not change everything at once. Move from familiar values to new values, then to different wording, then perhaps to a diagram or reverse question. If the learner fails at one transition, return to the relationship and explain the bridge. This is more informative than presenting a highly novel multi-topic question and concluding that the entire concept is missing. Unfamiliarity can contain several distinct demands.

Templates can have a legitimate role. A paragraph frame can help a novice organise an explanation. A standard layout can reduce unnecessary notation errors. The problem arises when the learner cannot tell whether the frame fits the task or cannot depart from it when needed. Teach the purpose of each part of a template. Then remove or vary parts when the learner has enough knowledge to make the corresponding decision.

Tricia, in an original English example, has learned to begin every explanation with “This shows that the character is determined.” On a new passage, the character repeatedly delays acting because they fear the consequences. Reusing determined produces a fluent but poorly supported response. The repair is not another opening phrase. Ask which behaviour the passage actually shows, what inference is justified and which detail supports it. Language should follow evidence.

A useful transfer discussion asks: what stayed the same, what changed, and why is this method still suitable—or no longer suitable? The explanation can be short. Students do not need an essay on every problem. A few well-chosen contrasts can reveal whether they are responding to the relationship or merely to a familiar word. The tutor can then choose the next example deliberately.

Be cautious about labels such as careless, rigid or unable to think. They describe the student broadly without specifying the teachable decision. “Uses the previous method whenever the word total appears” is more useful. It suggests a contrast set in which total appears in different relationships. The problem becomes something instruction can address instead of a character flaw the child is expected to overcome through willpower.

The existing mixed-practice guide and transfer guide remain the deeper teaching routes. This page uses those distinctions to answer a narrower parent question: why can substantial practice fail to improve the part of performance that the next school paper actually requires?

8. English difficulties that should not receive the same intervention

“Weak in English” can refer to reading accuracy, vocabulary, comprehension, inference, writing organisation, sentence control, editing or oral expression. These areas interact, but they are not interchangeable. A learner who cannot understand the passage needs different support from one who understands it but writes an answer that exceeds the evidence. Another may have clear ideas but struggle to organise them within the time available.

Start with a short, suitable passage and a small number of questions. Ask the learner to explain the situation in ordinary language. If the basic meaning is inaccessible, inspect unfamiliar words, sentence structure and reading. If the meaning is understood, inspect the question and the evidence selected. Avoid assigning another full comprehension paper before knowing which part of the process is failing.

An original passage and a bounded inference

Tricia placed the invitation beside her pencil case. At recess, she walked towards the music room, stopped at the door and checked the invitation again. She could hear the others laughing inside. After a moment, she straightened the folded corner of the paper and knocked.

A question asks, “What suggests that Tricia hesitated before entering?” A relevant answer points to stopping at the door and checking the invitation again. A reasonable explanation is that she was uncertain about entering. The passage does not establish that the others disliked her, that she had failed an audition or that she was being bullied. Those may be imagined stories, but they are not warranted conclusions from this short text.

If the student gives an unsupported story, ask which words support it. If no words do, help them narrow the claim. This is an evidence problem rather than simply a vocabulary problem. A more sophisticated adjective will not rescue an inference that goes beyond the text. The next practice should use another short passage in which several interpretations are possible but only some are supported.

If the student cannot understand invitation, recess or folded, vocabulary may be limiting access. Teach the relevant meanings in context, then return to the passage. Do not treat every unknown word as a reason to abandon comprehension until an enormous word list is mastered. Teach enough language to restore access while building vocabulary systematically over time. The Vocabulary Learning Hub provides broader routes without turning this tuition review into a second vocabulary master page.

Writing: improving the script is not the same as teaching the writer

Suppose a student writes, “I was angry. My friend was late. We missed the bus.” A tutor can rewrite this into an elegant paragraph, but the improved page alone does not show what the student learned. Choose a teachable target. Perhaps the learner needs to show the connection between the delay and the missed bus. Ask them to add one sentence explaining that relationship, then use the same principle in a different situation.

For a more advanced writer, the target may be the role of evidence in an argument. “Public transport is important because many people use it” is a claim with a broad reason, but it needs development. Ask what problem the service solves, for whom, under what conditions and with what limitation. The student should generate the reasoning. A supplied paragraph may be a useful model, but a fresh paragraph is needed to see whether the writer can use the structure.

Separate planning from sentence correction. A composition with a confusing sequence may not improve much from editing articles and tenses alone. A coherent plan may still produce difficult-to-read sentences. Review each layer in turn, then integrate. This prevents feedback from becoming a long list of unrelated faults. The student should know which one or two changes matter most in the next attempt.

EEF’s feedback guidance emphasises information that helps learners improve and opportunities to act on it. In practice, that means the student needs a response task after receiving comments. The precise revision method below is an illustrative teaching choice, not an EEF-prescribed protocol. Read the feedback evidence summary.

A practical writing loop is: identify one target, inspect a model, revise a small section, explain the revision and write a fresh section. For example, work on a clear topic sentence in one paragraph, then ask for a topic sentence on a different prompt. Later, check whether the skill appears in a complete piece without a reminder. This gives the tutor evidence at component and whole-task levels.

Do not insist that every learner must use impressive vocabulary to show progress. Precision, relevance and control may be the immediate priorities. A clear ordinary word can be stronger than an ambitious word used incorrectly. The parent question is not “How many advanced words were taught?” but “Which part of reading or writing can the child now manage more accurately, and where does the next difficulty begin?”

9. Science: distinguish remembering a term from reasoning with evidence

A Science response can contain relevant vocabulary and still fail to answer the question. The learner may recall a term without identifying the variable that changed, may describe an observation without explaining it, or may import a familiar explanation into an experiment that tests something different. Tuition should inspect the reasoning process rather than assume every lost mark can be repaired by memorising a longer model answer.

Begin by separating four tasks: state what was observed, identify a relationship in the data, propose an explanation supported by taught science, and evaluate what the experiment can establish. These tasks sometimes appear together, but the question may ask for only one. A student who writes an explanation when asked for an observation may be knowledgeable yet unresponsive to the task.

An original data example

Imagine a classroom activity comparing how far three toy cars travel after leaving the same ramp. Car A travels 80 cm, car B travels 95 cm and car C travels 70 cm in one trial. A question asks which car travelled furthest in that trial. The answer is B. A different question asks whether B will always travel furthest. The single trial does not establish that. Repeated trials and control of relevant conditions would be needed to support a broader conclusion.

The key teaching distinction is between the result observed and the general claim made. A student who writes “B is the best car” has not defined best and has moved beyond the measured outcome. A student who writes “B travelled furthest in this trial” stays within the evidence. This is not a matter of adding a keyword. It is a matter of matching the strength of a conclusion to the information available.

Now suppose the purpose is to compare ramp surfaces, but the experiment changes both the surface and the release height. The student should notice that the comparison cannot isolate the surface effect cleanly. Ask which conditions need to be kept comparable for the question being investigated. Do not demand a memorised list of every possible variable. Focus on the variables relevant to this design.

For explanation questions, ask the student to connect a change to an outcome through a taught relationship. If the link is missing, another round of vocabulary copying may not help. A tutor can model one causal chain, then ask the learner to build a different chain from familiar content. The example should remain within what the student has been taught; a novel apparatus should not also require untaught scientific knowledge.

For numerical Science, separate formula recall from formula selection and unit handling. A learner may know a relationship but substitute the wrong quantity. Another may substitute correctly but mix units. Marking only the final answer hides these differences. Ask the student to name each quantity before substitution and inspect the first line where meaning or units are lost.

Keep the teaching sequence proportionate. One targeted comparison may reveal the issue more clearly than a full paper. After repair, however, return the skill to an authentic question so it does not remain isolated. A learner who can identify a controlled variable in a simplified conversation should later recognise the same design issue in a written problem with diagrams or additional information.

Parents do not need to know every model answer to support this review. Ask the tutor, “Was the difficulty knowledge, reading the question, selecting the evidence, explaining the mechanism or writing the response?” The answer should point towards a teachable action. If the tutor says all are weak, ask which is the first priority and how that priority was identified.

One caution matters: a child may give a scientifically reasonable answer that does not match a parent’s remembered wording. Do not correct it solely because it sounds different. Use the school’s current expectations and an educator’s judgement where needed. The review should improve scientific understanding, not turn home into a competition between several adults’ preferred phrases.

10. Find where examination time is being spent

“Too slow” is an observation, not yet a diagnosis. Time can be spent reading, retrieving a fact, choosing a method, carrying out calculations, writing, checking or recovering from a wrong start. The same total duration can arise from very different processes. Before assigning more timed papers, examine a small task and identify the first substantial delay.

In an illustrative lesson, Kai Kai spends two minutes deciding how to start an equation problem, one minute solving it and two minutes checking the answer repeatedly. Telling him to calculate faster misses most of the time. A useful review might first address method recognition, then teach a bounded check. Another student who starts immediately but spends several minutes on basic arithmetic needs a different intervention.

Record a few landmarks rather than timing every movement. Note when the student begins a meaningful method, completes the main work and finishes checking. The record is approximate and local; it is not a psychological measure of processing speed. It simply helps tutor and student see where an otherwise vague time problem appears.

Protect untimed understanding. A learner who does not know the method needs instruction before pressure. A learner who is accurate without a clock but becomes disorganised with one may need a gradual progression: a generous time window, a short familiar set, a mixed section and later a full paper. The precise sequence should match the student’s needs and any approved arrangements.

Do not infer an anxiety disorder from a poor timed attempt. The student may be unfamiliar with the format, uncertain about the content, tired or responding to an unrealistic limit. If distress is persistent, significant or affects broader functioning, involve the school and appropriate qualified support. Academic observation can inform that conversation, but it is not a clinical diagnosis.

A paper’s chronology can matter more than the final cluster of errors. Suppose the student spends too long on an early difficult question, then rushes several accessible questions at the end. The late errors may be described as careless, but the first useful repair could be a decision about when to move on. Review the sequence before prescribing more accuracy drills.

Teach a move-on decision in advance rather than improvising it in an examination. The student can practise recognising when no new progress has been made, marking the place and returning later. The exact time allowance depends on the paper, marks and student’s circumstances; there is no universal number of seconds that suits every question. The important skill is making the decision deliberately instead of allowing one item to consume the rest of the paper.

Checking also needs a purpose. For an equation, substituting the result can test whether it satisfies the original statement. For an answer with units, checking that the requested quantity and unit match can catch an obvious mismatch. Reworking every question from the beginning may be unnecessary and expensive. Teach checks tied to likely errors, then observe whether the student uses them effectively.

A time improvement should not be celebrated without looking at accuracy. Finishing two minutes faster by guessing half the answers is not the intended gain. Conversely, taking slightly longer while removing several serious errors may be progress at an early stage. Record both time and outcome, then decide what balance the actual assessment requires.

The existing examination-proof guide provides a deeper route for moving from component skill to realistic performance. Here the parent decision is straightforward: before buying another paper package, find out whether the problem is knowledge, selection, execution, checking or paper management. The appropriate practice should follow that answer.

11. Homework and correction must have a clear job

Homework can consolidate teaching, create retrieval opportunities, reveal misunderstanding or prepare for the next lesson. It can also become an unexamined quantity that expands because completed pages are easy to see. Ask what the current assignment is supposed to change. The answer should be more specific than “practice is important.”

A student who needs fluent calculation may benefit from several carefully selected repetitions. A student who can calculate but chooses the wrong relationship may need a shorter mixed set. A writer who cannot develop a reason may need one well-supported paragraph rather than another whole essay. Different assignments can be equally serious because seriousness lies in the learning demand, not page count.

Set an expected scope and a response to difficulty. If a task is intended as a brief independent return, the family should know what to do when it becomes a long parent-led lesson. The student can mark the first inaccessible item, record the attempt and bring it back. This is useful feedback to the tutor. Quietly completing it together hides the mismatch and may lead the next assignment to become even harder.

Correction should identify what changes next. “Careless” does not tell a child how to prevent a repeated sign error. “The negative multiplier must apply to both terms in the bracket” does. After the explanation, ask for a fresh related item. Copying the corrected solution can be part of review, but it is not enough evidence that the student can now make the decision independently.

Use an active error record rather than a permanent catalogue of failure. An entry can contain the situation, the first wrong decision, a prevention cue and a later example showing whether the cue helped. When the problem stops recurring under suitable conditions, move it out of active attention. The archive may remain, but the learner should not revise every historical mistake every week.

Do not mark all errors as equally urgent. A one-off copying slip and a repeated misconception require different responses. An incorrect answer caused by an unknown concept deserves teaching. An answer produced through a known method with a minor execution slip may need a check or a short practice return. Prioritise the errors that are repeated, consequential or blocking current learning.

Homework review can be efficient. The tutor does not need to discuss every correct item. Sample the reasoning, inspect recurring losses and choose the next task accordingly. A programme that assigns work but never uses the responses loses a major source of information. A programme that spends the entire lesson remarking routine answers may use expensive instructional time poorly. The right balance depends on the learner and assignment.

Protect authentic work. For school assignments, follow the school’s rules about assistance, collaboration and tools. Parents and tutors can clarify, model on a separate example and give permitted feedback, but they should not produce a submission that falsely represents the student’s independent work. Honest evidence helps the school teach. A flawless page completed by several adults can be less educationally useful than an imperfect page that reveals a specific need.

The practical parent question is: “What is this homework for, how much independent work is reasonable, and what should happen when it reveals a gap?” A clear answer prevents much unnecessary conflict. It also lets the student learn that an unsuccessful attempt can be valuable when it leads to better instruction, rather than something that must be hidden behind completed pages.

12. A good teacher can still be in the wrong arrangement for this learner

Quality and fit are related but different. A clear lecturer may suit a student who needs organised revision and can practise independently. A learner with a persistent misconception may need more individual questioning. A small group may provide useful peer explanations, or it may move at a pace that leaves one child behind. The review should examine the actual teaching arrangement rather than assume that one format is universally best.

In one-to-one tuition, the opportunity is precision. The tutor can inspect a line of reasoning, change the example and respond immediately. The risk is that the adult does too much because help is always available. A useful individual lesson therefore includes deliberate opportunities for the learner to work without continuous commentary, followed by feedback. Personalisation should not mean uninterrupted prompting.

In a three-student illustrative class, Alicia may need help representing the problem, Tricia may choose a method correctly but explain it imprecisely, and Kai Kai may execute it slowly. One shared demonstration can still be useful, but their next tasks need not be identical. Each student can work on a different aspect and later compare reasoning. The group succeeds when the common teaching has a real shared purpose and individual differences remain visible.

EEF’s summaries describe both one-to-one and small-group tuition as potentially useful and emphasise targeted support and links with classroom learning. The summaries also report variation across studies, so they do not justify a guarantee based on group size alone. One-to-one evidence and small-group evidence are useful starting points for questions, not rankings of local providers.

For larger groups, ask how a quiet learner’s misunderstanding becomes visible. There may be written checks, consultation periods, submitted work or other feedback channels. A student who rarely volunteers can still be learning well, but the provider needs some way to know. Attendance and attentive appearance do not establish understanding. The mechanism for noticing individual difficulty matters more than a reassuring claim that all students are monitored.

For online tuition, inspect the work rather than the platform’s appearance. Can the tutor see the student’s handwritten reasoning? Does the student solve or write during the lesson? Are there bounded independent intervals? Is the learner able to ask for clarification? A screen-shared model answer may explain clearly, but the programme still needs evidence of student generation. Saving travel is valuable only if the learning arrangement remains useful.

For recorded lessons, ask whether the learner pauses to attempt tasks and has a route for unresolved questions. A highly independent student may use recorded teaching efficiently. Another may watch fluently without producing any work. The difference does not prove that recorded teaching is good or bad in general. It identifies whether the format supplies the functions this student currently needs.

Fit includes timing and continuity. A lesson placed after an unusually demanding school day may be difficult to use even when the teaching is strong. Repeated tutor changes may require the student to rebuild a shared understanding of methods and expectations. These are service-design issues worth discussing. They should not automatically be interpreted as a lack of effort by the child.

Keep the provider conversation specific. Instead of “The class is not personalised,” ask, “How is the recurring equation-translation problem being checked, and what does the student do after the shared explanation?” Instead of “The group is too big,” ask, “How does the tutor identify what my child can do independently?” Answers to those questions are more actionable than format labels alone.