The Tutor Handbook Vol No.0038

The Tutor Handbook · Volume 0038 · Series ID THB-0038

A tutor can choose the right intervention and still give too much of it.

That is the problem of dose.

A good explanation can become an over-explanation. A useful hint can become a chain of hints. A worked example can become a permanent crutch. A study plan can become a timetable the student never learns to build. A careful tutor can become so responsive that the learner slowly stops needing to decide what to do next.

The danger is subtle because the lesson may look excellent while it is happening.

The student answers. The page fills. The tutor is active. The parent hears that the child completed difficult work. Everyone can point to visible progress.

Then the tutor leaves the system.

The student meets a fresh question, an unfamiliar passage, a mixed Mathematics paper, an examination clock, a blank composition page, or an evening study session with nobody sitting beside them.

Now the real question appears:

How much of the successful performance belonged to the learner, and how much belonged to the support?

The previous chapter, The Tutor Handbook Vol No.0037 | The Branch Point, asked which intervention should be chosen after the cause has been separated. This chapter asks the next question: once the right branch has been selected, how much help should be delivered, how quickly, how specifically, for how long, and with what withdrawal plan?

This chapter sits inside the wider eduKate Sengkang learning architecture. How Learning Works maps durable capability change. How Studying Works owns deliberate learner activity outside direct teaching. eduKate Sengkang Education Runtime connects learner state, diagnosis, repair, practice, transfer, performance and independence. How to Improve Anything provides the larger improvement loop. The Tutor Classification Model by eduKateSG remains the tutor ladder from Class 0 Homework Helper to Class 6 Learning Architect.

Quick Read

The branch chooses what kind of help to give. The dose chooses how much, when, how specifically, how often and for how long.
Good support is contingent on the learner’s present state, not fixed by tutor habit.
The smallest successful dose is not always the shortest explanation. It is the least external control that produces useful independent change.
Help given too early can erase diagnostic evidence.
Help given too late can turn challenge into unproductive failure.
A support dose should preserve as much learner thinking as the learner can presently carry.
Every dose needs a success signal, an escalation rule, a fading rule and a stop rule.
Support should become lighter as capability becomes stronger.
If performance collapses the moment support is removed, the dose created assisted performance, not yet independent capability.
Different learner states need different doses even when the topic is identical.
The best dose changes across learning, studying, training, education and examination performance.
The tutor’s final job is not to become permanently useful. It is to make the right kind of help progressively less necessary.

1. What “Dose” Means in Tutoring

“Dose” is a metaphor for instructional intensity. It does not turn teaching into medicine, and it does not imply that students are patients. It is simply a useful way to force precision.

A tutor does not only choose an intervention. The tutor also chooses how much of that intervention enters the learner’s system.

One question or five questions?
One cue or a full worked solution?
One model or four?
Five repetitions or fifty?
A temporary checklist or a permanent template?
Thirty seconds of wait time or immediate rescue?
One week of structured planning or an adult-managed timetable all term?

The content may be the same. The dose is different.

2. Branch and Dose Are Different Decisions

The Branch Point asks: is the learner missing a concept, a representation, a retrieval route, a discrimination rule, a fluent procedure, a study routine, a transfer condition, or performance control?

The Dose asks: now that we know the likely mechanism, how much support is necessary before the support itself begins doing work that should belong to the learner?

A tutor can therefore make four different mistakes:

right branch, right dose;
right branch, wrong dose;
wrong branch, apparently helpful dose;
wrong branch, excessive dose that hides the error completely.

The second and fourth mistakes are especially difficult to notice because the student may improve during the lesson.

3. The Assistance Dilemma

Educational researchers have long studied a tension sometimes described as the assistance dilemma: too little help can leave learners stuck, while too much help can reduce the thinking, search, retrieval and decision-making that produce learning.

The tutor therefore has two responsibilities that pull in opposite directions.

Prevent wasteful, repeated or damaging failure.
Preserve enough learner work for capability to actually grow.

The art is not choosing help or no help.

The art is choosing the right amount of help at the right moment for the right job.

4. The Six Dimensions of a Support Dose

Amount: how much information or structure is supplied?
Timing: when does the tutor intervene?
Specificity: how close does the help come to the answer or next step?
Frequency: how often is support supplied?
Duration: how long does the support remain available?
Ownership: how much of the cognitive work remains with the learner?

Two tutors can say exactly the same sentence and still deliver different doses because one says it after the learner has attempted, while the other says it before the learner has even read the question.

Timing changes dose.

5. Amount: How Much Structure Enters the Task?

The first dimension is the most obvious.

A tutor can point to a line, ask a question, name the concept, supply a formula, provide a partial example, solve the first step, solve half the problem, or demonstrate the entire solution.

Each move transfers a different amount of structure into the task.

The useful question is:

What is the smallest amount of added structure that allows the learner to perform the missing cognitive operation?

6. Timing: Help Before, During or After the Attempt?

Help before an attempt changes the task the learner receives.

Help during an attempt changes the route while the learner is travelling.

Help after an attempt turns the learner’s own work into evidence.

The same explanation can therefore teach three different things depending on when it arrives.

Before: “Here is the method.”
During: “You are at the decision point; what condition determines the method?”
After: “Compare your chosen method with the condition in the question. Where did the route diverge?”

When possible, preserve enough of the learner’s first attempt to reveal what the learner can already do.

7. Specificity: How Close Does the Help Come to the Answer?

“Check the relationship” is less specific than “use simultaneous equations.”

“Look at the evidence in the second paragraph” is less specific than “the answer is in the sentence beginning ‘Although’.”

“What quantity is the percentage based on?” is less specific than “the base is the original amount.”

Specificity should rise only when broader prompts fail for a useful reason.

If the tutor jumps immediately to a highly specific cue, the student may never practise recognising the condition independently.

8. Frequency: One Prompt Is Different From Continuous Prompting

A single cue can restart a stalled process.

A cue before every step can become the process.

This is one of the easiest ways for good tutoring to create dependence. The tutor feels helpful because every hesitation receives support. The learner feels comfortable because uncertainty is quickly removed. But the learner never becomes responsible for deciding when to continue, reconsider or check.

Frequency is therefore not merely how often help is offered. It is how often the learner is allowed to experience and resolve uncertainty before help arrives.

9. Duration: A Temporary Scaffold Can Quietly Become Permanent

A checklist used for three lessons may be an excellent scaffold.

The same checklist used for six months without any attempt to shorten it may be a warning.

Support that remains unchanged while learner capability changes is no longer calibrated support. It has become background infrastructure.

Every support should therefore have a planned review date.

10. Ownership: Who Is Doing the Cognitive Work?

The most important dimension is ownership.

Imagine a comprehension question. The tutor reads the passage, identifies the relevant sentence, explains the inference, tells the learner the answer boundary, and asks the student to phrase it.

The learner may write a good answer.

But most of the difficult cognitive work has already been done.

Now imagine the tutor says only, “Show me the sentence that makes your inference safe.” The learner must search, select, justify and then write.

The final answer may look similar. The ownership is completely different.

11. The Minimum Sufficient Dose

The core rule of this volume is:

Give the smallest amount of support that changes the limiting mechanism while preserving the maximum useful amount of learner thinking.

This is not a demand to make learning artificially hard.

It is a demand to keep the tutor from performing work that the learner is ready to perform.

12. Minimum Does Not Mean Weak

A minimum sufficient dose can be powerful.

One carefully chosen contrast may expose a misconception more effectively than a twenty-minute explanation. One changed-surface question may reveal transfer failure more clearly than a whole worksheet. One demand for a reason may expose whether a student is choosing a method or merely recognising a familiar page.

Small support can create large thinking.

13. The Under-Dose

Too little support has its own cost.

The learner repeats the same misconception.
The task becomes random guessing.
Working memory is consumed by irrelevant complexity.
The learner cannot see the relationship that practice is supposed to strengthen.
Confidence erodes because failure supplies no useful information.

Productive independence is not abandonment.

If the learner cannot yet perform the core process, refusing support does not magically create capability.

14. The Over-Dose

Too much support creates a different set of problems.

The tutor names the method before the learner selects it.
The tutor supplies the evidence before the learner searches.
The tutor corrects the line before the learner checks.
The tutor turns every question into a guided example.
The tutor manages every study decision.
The learner becomes fluent at following rather than initiating.

Over-support can produce smooth lessons and fragile independence.

15. The False Fluency Problem

Support can make performance look fluent before learning is fluent.

A learner answers quickly because the tutor has already narrowed the decision space. A student writes smoothly because the paragraph architecture is supplied. A Mathematics solution flows because the method has been announced. A Science answer looks precise because the tutor has already separated observation, mechanism and consequence.

None of this is useless. Guided performance can be an essential stage.

The mistake is calling it independent mastery before the support is removed.

16. Assisted Performance Is a State, Not a Failure

A learner who succeeds with help is not “faking it.” Assisted performance is useful evidence.

It tells the tutor how close the learner is.

Success after a broad cue suggests more ownership than success after a full model.
Success after a single prompt suggests more independence than success after prompts on every step.
Success after one worked example followed by independent transfer suggests more learning than success only while examples remain visible.

The job is to measure the distance between assisted and independent performance, then shrink it.

17. The Support Ceiling

Every task has a point beyond which more help begins replacing the target skill.

If the target is inference, supplying the inference crosses the ceiling.

If the target is method selection, naming the method crosses the ceiling.

If the target is planning, giving the complete plan crosses the ceiling.

If the target is error checking, identifying every error crosses the ceiling.

The support ceiling depends on the learning objective. Help that is harmless in one lesson may destroy the measurement in another.

18. The Tutor Should Know What Must Remain With the Learner

Before giving help, state the protected cognitive job.

“The learner must still choose the method.”
“The learner must still retrieve the definition.”
“The learner must still find the evidence.”
“The learner must still decide what to write first.”
“The learner must still notice the contradiction.”

This one habit prevents many accidental over-doses.

19. Wait Is a Real Intervention

Many tutors underestimate silence.

The learner pauses. The tutor interprets the pause as failure. A hint arrives. The learner continues.

But hesitation may be retrieval, comparison, error detection or planning in progress.

If the tutor always rescues the first visible pause, the learner never practises working through uncertainty.

Waiting is not passive when the tutor is observing what the learner does next.

20. The Rescue Threshold

Wait should not become a ritual.

A tutor needs a rescue threshold: a point at which continued struggle is unlikely to produce useful information or useful learning.

The learner is repeating the same wrong operation.
The learner has lost the task entirely.
The learner is guessing without a model of what to inspect.
The error is about to contaminate many downstream steps.
The emotional cost is rising while informational value is falling.

At that point, intervene.

21. Productive Struggle Is Not Maximum Struggle

“Let them struggle” is not a sufficient teaching principle.

Struggle is useful when the learner has enough structure to search, enough prior knowledge to test possibilities, and enough feedback to update the route.

Struggle becomes unproductive when the student is simply accumulating failed attempts without learning what the failures mean.

The tutor should preserve challenge, not worship difficulty.

22. The First Dose Is Usually Diagnostic

Start with help that teaches the tutor something.

A broad prompt is valuable because the learner’s response reveals whether the missing piece is attention, retrieval, concept, method, representation or execution.

If the first prompt is already the answer, the tutor loses that information.

Good first doses are often reversible, low-information and high-diagnostic-value.

23. The Escalation Ladder

Observe.
Wait.
Ask the learner to restate the task.
Direct attention.
Offer a broad cue.
Offer a discriminating question.
Provide a partial structure.
Model one decision.
Model the process.
Guide the learner through a fresh attempt.
Rebuild the prerequisite or route if the evidence now requires it.

This is not a rigid staircase. Some learners should begin higher because the missing structure is already known. Some tasks are too time-sensitive for a slow climb.

But the ladder gives the tutor a discipline: do not jump to maximum help simply because maximum help is easy to deliver.

24. The De-Escalation Ladder

Support should also have a path downward.

Full model.
Partial model.
Prompted attempt.
Broad cue only.
Wait and observe.
Independent attempt.
Independent attempt on a changed surface.
Independent attempt after delay.
Independent attempt under authentic performance conditions.

If the tutor can escalate but cannot de-escalate, the tutoring system only knows how to add.

Good tutoring must know how to remove.

25. Fading Is Not Abrupt Withdrawal

A learner who succeeds with a scaffold may still fail if the scaffold disappears all at once.

Fading means reducing external structure while increasing learner responsibility.

Complete worked example → missing final step.
Missing final step → missing two middle steps.
Partial example → headings only.
Headings → one cue.
One cue → independent reconstruction.

The learner should feel the cognitive load transferring gradually back into their own control.

26. Fading Must Follow Evidence

Do not fade support because three lessons have passed.

Fade because the learner has shown readiness.

The learner anticipates the next step.
The learner self-corrects before the tutor speaks.
The learner uses the scaffold without tutor reminders.
The learner succeeds after a lighter cue.
The learner can explain the decision rule.
The learner can recover from a small error independently.

Fading should be response-sensitive, not calendar-sensitive.

27. Expertise Changes the Correct Dose

Support that helps a novice can become redundant for a more experienced learner.

This matters because tutors often reuse successful teaching routines after the learner has outgrown them.

The student who once needed every algebraic transformation written explicitly may later need only a check at high-risk transitions. The writer who once needed a paragraph frame may later need only a planning question. The reader who once needed evidence highlighted may later need uncertainty left intact so evidence selection can be trained.

Same learner. Same subject. New dose.

28. Novice Dose

Novices often need more visible structure because they cannot yet tell which information matters.

clear examples;
explicit relations;
guided comparison;
limited choices;
immediate correction of foundational misconceptions;
short independent attempts between supported examples.

The goal is not to keep the novice comfortable. It is to help them build a usable internal model quickly enough that support can begin shrinking.

29. Intermediate Dose

Intermediate learners often need less explanation and more decision-making.

mixed examples instead of labelled blocks;
partial worked solutions instead of complete models;
questions that ask why a method applies;
changed surfaces;
delayed retrieval;
support only at the point where the learner’s process actually breaks.

The tutor’s voice becomes quieter because the learner’s decision system should become louder.

30. Advanced Dose

Strong learners can be over-taught too.

An advanced student may need less explanation, fewer routine questions, higher-quality feedback, harder discrimination, more authentic constraints and longer periods of independent problem solving.

For strong learners, the tutor’s best contribution may be to create better problems, notice smaller errors, ask more exacting questions and refuse to interrupt a productive line of thought.

31. Confidence Does Not Automatically Mean “Give More Help”

A learner who lacks confidence may ask for help earlier than the task requires.

The caring response is not always immediate rescue.

Sometimes the tutor should reduce the size of the next step, confirm what is already correct, or ask the learner to choose between two plausible routes instead of supplying the route.

Confidence grows when the learner accumulates evidence of successful control, not merely evidence that help is available.

32. Urgency Changes the Dose

A learner six months from an examination and a learner two days from an examination may need different support.

With time, the tutor can preserve more discovery, retrieval and independent reconstruction.

Under genuine short-term pressure, a stronger scaffold may be justified to protect performance.

But temporary emergency support should be labelled temporary. After the immediate demand passes, the tutor should return to the deeper learning problem.

33. Time Pressure Can Create Hidden Over-Dosing

Tutors often over-help because they want to finish the planned lesson.

The student is slow. The worksheet is long. The clock is moving. The tutor starts completing the learner’s thinking to keep the lesson on schedule.

The lesson plan gets completed. The learning objective may not.

If necessary, complete fewer questions and preserve the cognitive work that matters.

34. The Learning Dose

When the target is durable learning, support should help the learner construct, retrieve, connect and transfer.

Model enough to reveal the relation.
Ask the learner to reconstruct it.
Provide feedback on the reconstruction.
Remove the example.
Test on a fresh item.
Change the surface.
Return after delay.

Immediate smoothness is not the main measure. What survives matters more.

35. The Studying Dose

When the target is studying, the tutor must avoid becoming the student’s executive function forever.

A tutor may initially help plan the week, prioritise tasks and schedule retrieval. But the dose should change.

Tutor-built plan.
Joint plan.
Student proposes; tutor checks.
Student plans independently; tutor samples.
Student reviews and adjusts their own plan.

The study system is only truly learned when the learner can operate it without the tutor managing every decision.

36. The Training Dose

Training dose concerns repetition, variation, difficulty and load.

Too little repetition and the process remains fragile. Too much identical repetition and the learner becomes efficient only inside one narrow condition.

The dose should evolve:

few accurate repetitions;
then varied repetitions;
then mixed selection;
then delayed return;
then realistic performance conditions.

37. The Education Dose

Education contains more than one lesson.

A learner may face curriculum transitions, subject combinations, different teacher expectations, examinations, school routines and growing demands for independence.

The tutor should not solve every system problem for the child. Instead, help the learner interpret the system, identify which demands are stable, choose routines that travel across subjects and gradually assume responsibility for navigating the environment.

38. The Performance Dose

Performance support should reproduce the pressure that the learner must eventually handle.

If the tutor stops the clock every time a learner hesitates, the session may train content but not timed performance.

If the tutor gives a hint during every difficult examination item, the learner never practises leaving, returning, allocating time or recovering after uncertainty.

The support dose must respect the performance variable being trained.

39. Class 0 · Homework Helper

The Homework Helper should use the smallest support that gets work moving without becoming the person who must be present for work to begin.

First organise materials.
Then ask the child to state the task.
Then ask for the first action.
Only then add more support if the first action cannot be identified.

Class 0 dosage is successful when homework needs less adult activation over time.

40. Class 1 · Explainer

The Explainer is especially vulnerable to over-dosing because explanation feels productive.

A clear tutor can keep explaining long after the learner has enough information to begin thinking.

The Explainer should therefore use a loop:

Explain one relation → ask the learner to reconstruct it → test it → explain again only where reconstruction fails.

The goal is not maximum clarity delivered by the tutor. It is maximum clarity usable by the learner.

41. Class 2 · Drill Builder

The Drill Builder controls repetition dose.

Five accurate, feedback-rich repetitions may be enough to stabilise a fragile step. Twenty may be justified if fluency remains weak. Fifty may be wasteful if the learner already performs accurately and now needs selection or transfer.

Good drill ends when the adaptation has occurred, not when the worksheet is empty.

42. Class 3 · Diagnostic Tutor

The Diagnostic Tutor uses dose partly as a probe.

If a broad cue works, retrieval or attention may be the issue.
If a cue fails but a contrast works, discrimination may be weak.
If contrast fails but a model works, the internal representation may be missing.
If modelling works only while visible, transfer or retrieval may remain weak.

Response to different doses is diagnostic evidence.

43. Class 4 · Route Designer

The Route Designer controls dose across time.

A route can start heavily scaffolded and progressively demand more independent work. A route can begin with blocked practice and gradually mix. A route can move from tutor-created retrieval schedules to learner-created schedules.

Class 4 asks whether the programme itself is producing decreasing dependence.

44. Class 5 · Performance Coach

The Performance Coach must resist rescuing the learner from the very pressure being trained.

The coach can debrief after the timed set, pause only at defined checkpoints, or gradually increase realism. But the final training environment must allow the learner to experience the clock, uncertainty, fatigue and recovery without live tutor steering.

45. Class 6 · Learning Architect

The Learning Architect asks whether support is being delivered at the wrong layer.

A child may appear to need more content explanation when the real constraint is study initiation. A learner may appear to need more practice when the real problem is method selection. A student may appear to need a confidence intervention when the real issue is repeated uncorrected foundational failure.

The highest-level dose decision can therefore be to reduce support in one layer and strengthen a different one.

46. Why Three-Student Tutorials Change the Dose Problem

In a three-student tutorial, the tutor is close enough to see working but not forced to sit inside one learner’s thinking every second.

That matters.

The tutor can observe one learner while another attempts independently. Students can see that uncertainty is normal. The tutor can delay rescue without abandoning the class. Peer pace can create useful comparison without turning the room into a large lecture.

The format does not automatically produce independence, but it makes calibrated attention possible.

47. One-to-One Can Accidentally Increase the Dose

One-to-one attention is valuable when the learner needs close observation, intensive repair or a quiet environment.

Its risk is continuous availability.

The tutor sees every hesitation and can answer every question immediately. If both parties are not careful, the learner starts outsourcing uncertainty management to the tutor.

Good one-to-one tutoring therefore needs deliberate silence, independent intervals, delayed help and tasks that the learner must complete without conversational steering.

48. Large Groups Can Under-Dose Visibility

A large group has the opposite risk.

The teacher may deliver an appropriate average dose while individual learners receive too little diagnosis, too little correction or support at the wrong moment.

The dosage problem is therefore not solved by class size alone. It depends on whether learner state can be observed and support can be adjusted.

49. Peer Help Has a Dose Too

A peer can explain, compare, question or simply give the answer.

Good peer help preserves the target thinking.

“What did you use to decide?” is usually a better learning prompt than “Do this formula.”

Tutors should teach students how to help one another without stealing the cognitive work.

50. Parent Support Has a Dose

Parents often begin helping because the child is genuinely struggling.

Then the help becomes routine.

Parent reminds the child to begin.
Parent checks every answer.
Parent manages the bag.
Parent reads every school message.
Parent decides the revision sequence.

Some of these may be appropriate temporarily. The question is whether the responsibility is transferring.

A useful parent-support rule is:

Support the system the child is learning to run; do not quietly become the system.

51. Tool Support Has a Dose

Calculators, model answers, formula sheets, grammar checkers, search engines and AI tools can all be useful.

The dosage question is the same:

Which cognitive job is the tool allowed to perform, and which job must remain with the learner?

If the target is idea generation, a tool that generates every idea removes the target. If the target is editing, a tool can provide a second-pass comparison after the student edits first. If the target is learning a method, a tool may generate practice while the student still chooses and executes the method.

Tools should reduce irrelevant friction without erasing the capability being trained.

52. Homework Dose

More homework is not automatically more learning.

The homework dose depends on purpose.

For retrieval: short repeated returns may beat one long block.
For fluency: enough repetitions to stabilise, then stop.
For transfer: fewer questions with meaningful variation may be more useful.
For performance: realistic timing and completion matter.
For diagnosis: one carefully chosen item may reveal more than twenty familiar ones.

Volume should follow mechanism.

53. Retrieval Dose

Retrieval needs enough difficulty to require recall but not so much failure that every attempt becomes answer exposure.

If a learner cannot retrieve, the tutor can supply a partial cue, then remove it on the next return. If retrieval is easy, increase delay, mix the material or ask for reconstruction in a new context.

The dose moves with memory strength.

54. Concept Dose

Concept teaching is often over-dosed through repeated explanation.

After one clear explanation, ask the learner to predict, compare, draw, explain or apply.

If the learner cannot, locate the missing relation and explain that part.

Conceptual dose should alternate between information entering and learner structure coming back out.

55. Method-Selection Dose

When method selection is the target, the tutor must protect the choice.

Do not put the chapter title above the question. Do not name the method in the instruction. Do not arrange practice so every item uses the same procedure.

If the learner stalls, cue the discriminating feature rather than the method itself.

This keeps the dose below the support ceiling.

56. Writing Dose

A writing tutor can easily overtake the page.

The tutor suggests the opening, improves the vocabulary, reorganises the paragraph, supplies the transition and rewrites the ending. The composition becomes better. The writer may not.

Protect specific writing jobs.

If idea generation is the target, do not supply the ideas.
If organisation is the target, let the learner sequence.
If sentence control is the target, feedback may be more specific.
If revision is the target, mark the location of weakness without rewriting the sentence first.

57. Reading and Comprehension Dose

Reading support should help the learner notice structure, evidence and inference boundaries without turning the tutor into a live answer key.

A useful progression is:

“What is the question asking?”
“Which part of the passage matters?”
“What evidence makes your answer safe?”
“What is the smallest answer that fully satisfies the question?”

If the learner can perform these moves independently, the tutor should stop supplying them.

58. Mathematics Dose

Mathematics tutoring often becomes over-guided line by line.

Instead, identify the decision points.

What is known?
What is unknown?
What relationship connects them?
Which representation makes that relationship visible?
Which operation follows?
Does the answer make sense?

The tutor may help at one decision point while leaving the rest intact.

59. Science Explanation Dose

For Science, the learner may need help connecting observation to scientific relation to consequence.

A full model may be useful once. Then reduce it.

First model: observation → mechanism → consequence.
Next question: headings only.
Next: ask which bridge is missing.
Next: independent answer.
Then: changed experiment or unfamiliar context.

The structure should migrate from the page into the learner’s reasoning.

60. Additional Mathematics Dose

Additional Mathematics often combines conceptual choice with algebraic execution.

If the calculus idea is correct but algebra fails, do not over-dose calculus explanation. If the algebra is secure but the student cannot choose the calculus tool, do not hide that selection problem by announcing the method.

The dose should enter exactly where the chain first becomes unreliable.

61. Examination Dose

Examination training eventually requires help withdrawal.

Early in training, the tutor may pause after a section, ask for a timing decision or remind the learner to recover from a difficult item.

Later, those prompts must disappear.

If the exam requires the learner to manage time, uncertainty, checking and recovery alone, the final training dose must leave those responsibilities with the learner.

62. Vocabulary Dose

A vocabulary learner can be over-supported by definitions, example sentences and answer options that remain visible during every retrieval attempt.

Support can fade from full meaning to category cue, from category cue to first-letter cue, from cue to independent retrieval, then to production in an unfamiliar sentence.

The objective is not recognition of the study page. It is usable language.

63. Alicia: One Prompt Too Many

Alicia reaches a difficult Mathematics question and pauses.

The tutor asks, “What quantity is fixed?” Alicia answers correctly. Before she can continue, the tutor asks, “So which equation will you use?”

The second prompt feels helpful.

It may be one prompt too many.

If recognising the fixed quantity was enough to reactivate the method, the tutor should wait. The next decision belongs to Alicia.

64. Beatrice: The Worksheet Is Already Too Helpful

Beatrice completes ten algebra questions accurately.

The worksheet title says “Simultaneous Equations.” Every item uses simultaneous equations. The worked example remains at the top of the page.

The tutor has said almost nothing, yet the support dose is still high because the environment has already removed method selection.

Dose is not only verbal assistance. Task design can over-help too.

65. Ciara: Timing Help Without Taking Over the Clock

Ciara understands comprehension but loses control late in timed work.

During early training, the tutor marks two checkpoints in the set. At each checkpoint, Ciara states whether she is ahead, on time or behind and chooses what to do next.

After several sessions, the checkpoints remain but the tutor says nothing.

Later, the marks disappear too.

The timing dose fades from external reminder to self-monitoring.

66. Denise: Correcting the Error Without Owning the Check

Denise makes a sign error.

The tutor could circle the sign. Instead, the tutor says, “One line changes the meaning. Find it.”

Denise checks, finds the line and repairs it.

On the next question, the tutor says nothing.

Denise checks the risky transition herself.

The correction taught checking because the dose preserved error detection.

67. Emily: Planning Support That Shrinks

Emily feels overwhelmed by six competing deadlines.

Week one, the tutor demonstrates a simple priority rule and builds the plan with her.

Week two, Emily ranks the tasks and the tutor checks.

Week three, Emily arrives with a plan and explains her choices.

Week four, the tutor samples only one decision.

The plan has become lighter because the planner has become stronger.

68. Faith: From Science Frame to Independent Explanation

Faith initially needs a visible three-part frame for Science explanations.

The tutor does not keep the frame forever.

It becomes three blank labels. Then one question: “What connects the observation to the result?” Then no prompt. Finally, Faith receives an unfamiliar investigation where the surface details have changed.

The dose succeeds when the reasoning structure survives after the visual structure disappears.

69. The Help-Request Problem

Students also control dose by how they ask for help.

“I don’t know” invites the tutor to decide what kind of help to give.

“I understand what the question wants, but I cannot decide whether this relationship is additive or multiplicative” is a much better request.

Strong help-seeking narrows the uncertainty before assistance enters.

70. Teach Learners to Ask for the Smallest Useful Help

“Can you tell me whether my first step is sensible?”
“Can you give me one clue, not the method?”
“Can you point me to the sentence I should reread?”
“Can you check my plan but let me write the paragraph?”
“Can you tell me which part is wrong without correcting it?”

A learner who can request calibrated help is becoming a stronger self-regulated learner.

71. The No-Help Test

After supported success, remove help and ask for a fresh attempt.

This is the simplest independence test.

If the learner succeeds, the support may have done its job.

If the learner immediately asks for the same cue, the support has not yet migrated into independent control.

Do not treat that as failure. It tells you the next dose.

72. The Fresh-Question Test

The learner should not only repeat the supported item.

Give a fresh question that requires the same mechanism.

If success disappears, the learner may have remembered the local route rather than learned the general relation.

Fresh questions prevent the tutor from confusing guided memory with independent capability.

73. The Changed-Surface Test

Change wording, numbers, representation, context or position while preserving the underlying structure.

Then observe how much support returns.

If the learner again needs the full model, the original dose may have created local success without transfer.

If only one broad cue is needed, transfer is developing.

74. The Delayed-Return Test

Immediate independence can still be inflated by recency.

Return after a day, a week or an appropriate interval.

Ask again with less support.

Durable learning reduces the amount of help required after time has passed.

75. The Authentic-Condition Test

A learner may need no support untimed and still need heavy support under pressure.

That does not necessarily reopen content teaching.

It may mean the dose must shift from instructional support to performance training: pacing, recovery, selection under pressure, working discipline, uncertainty management and checking.

76. Self-Correction Is a Powerful Dose Signal

When learners begin correcting themselves before feedback arrives, the tutor should notice.

Self-correction means part of the feedback loop has moved inside the learner.

At that point, external checking should become less frequent and more selective.

If the tutor continues correcting every small error first, the tutor may suppress the very self-monitoring that is beginning to emerge.

77. The Dose Ledger

For a significant intervention, record:

target mechanism;
protected learner job;
starting support dose;
success signal;
under-dose signal;
over-dose signal;
next lighter dose;
next stronger dose;
fresh-question result;
changed-surface result;
delayed-return result;
authentic-condition result;
stop condition.

The ledger prevents support from continuing merely because it has become familiar.

78. The Under-Dose Signal

Increase support when:

the learner cannot begin because a prerequisite relation is missing;
the same misconception is being rehearsed;
the task is generating guesses rather than reasoned attempts;
the learner cannot use feedback because they do not understand the underlying distinction;
the informational value of more independent failure has become very low.

More support is justified when it restores useful thinking.

79. The Over-Dose Signal

Reduce support when:

the learner waits for prompts that are no longer necessary;
the tutor speaks before the learner attempts;
the learner performs only when the same scaffold is present;
the tutor cannot tell what the learner would do alone;
the task has become easier because decision points have been removed;
the learner can explain what to do but is not being allowed to do it independently.

Over-dose is not measured by how much the tutor talks. It is measured by how much learner control the support displaces.

80. The Stop Rule

Stop the intervention when the learner can perform the target process independently across the conditions that matter.

fresh item;
changed surface;
appropriate delay;
realistic level of load;
reasonable pressure;
without the original scaffold.

Do not keep a successful intervention alive merely because it was successful.

81. The Re-Dose Rule

Previously removed support may occasionally need to return.

A harder context, long delay, new curriculum demand or pressure spike may expose fragility.

Re-dose lightly first.

Do not automatically restore the entire old scaffold. Use the smallest reminder that reactivates the internal process, then withdraw again.

82. The Plateau Can Be a Dose Problem

A learner can plateau because support is too low.

A learner can also plateau because support is too high.

If every difficult decision is pre-structured, the learner may never develop the next level of selection, planning, monitoring or transfer.

When progress stalls, inspect not only what is being taught but how much of the task the tutor is carrying.

83. Transfer Failure Can Reveal Over-Support

If performance is strong inside tutoring but weak outside, ask whether the tutoring environment contains invisible supports.

The tutor always identifies the topic.
The worksheet always groups similar questions.
The student always receives immediate feedback.
The tutor always notices errors before the student does.
The study plan always comes from an adult.

The learner may know more than before and still be undertrained for independent deployment.

84. Misconception and Cognitive Load Need Different Doses

A misconception may require direct conceptual correction.

Excessive task complexity may require simplification without changing the concept.

Those are different interventions and different doses.

If a learner understands the principle but is overloaded by a crowded diagram, simplifying the representation may be enough. If the learner holds the wrong principle, simplifying the diagram will not repair the misconception.

85. Multiple Bottlenecks Need Sequential Dosing

A complex task can fail in several places.

Do not attack all weaknesses at once unless the context demands it.

Repair the controlling bottleneck, retest the system, then see what becomes visible next.

Large multi-part interventions can make improvement impossible to attribute. Sequential dosing preserves information.

86. Motivation Support Has a Dose

Encouragement is useful, but endless encouragement cannot replace a working route.

If the learner is discouraged because they repeatedly fail without understanding why, the most motivating intervention may be a precise diagnosis and one solvable next step.

Use emotional support to stabilise engagement, then return agency to the learner through evidence of capability.

87. The Tutor Should Not Diagnose Beyond the Tutor’s Role

The Dose is an educational operating model, not a medical or psychological diagnostic system.

If a learner shows serious distress, health concerns, safety risk or needs beyond the tutor’s professional role, the correct response is not to keep adjusting educational dose indefinitely. Involve the appropriate parent, school professional or qualified specialist.

88. Parent Communication About Dose

Parents can mistake less visible tutor intervention for less teaching.

Explain the purpose clearly:

We are deliberately reducing prompts because your child can now perform more of the process independently. We will still intervene if the same mechanism breaks, but the next stage is to make sure the improvement belongs to the learner.

That turns silence from apparent neglect into visible pedagogy.

89. Student Communication About Dose

Students should know why support is changing.

“I am going to give you less information this time because the last two attempts were strong.”
“I will tell you where the issue is, but I want you to repair it.”
“I am not naming the method because choosing it is today’s skill.”
“This set is timed, so I will not rescue you during the question. We will debrief after.”

When learners understand the dosage logic, reduced help feels like progression rather than withdrawal.

90. The Ninety-Second Dose Check

Before helping, a tutor can run six questions quickly:

What is the learner trying to do?
What part can they already do?
Where does the process first become unreliable?
What is the smallest support that targets that point?
What thinking must remain with the learner?
What will I remove next if this works?

This prevents help from becoming reflexive.

91. The Ninety-Minute Lesson Dose Pattern

A strong tutorial does not need the same support intensity for ninety minutes.

A useful pattern may look like this:

Opening: low support, observe recall and current state.
Teaching: enough structure to make the new relation visible.
Guided practice: moderate support at defined decision points.
Independent practice: reduce prompts sharply.
Transfer: change the surface and preserve uncertainty.
Review: ask the learner to explain what now belongs to them.
Handover: define what they will do without the tutor.

The dose should usually decrease as the lesson moves from introduction to ownership.

92. The Weekly Dose Review

Once a week, ask:

Which prompts are disappearing?
Which scaffolds are shrinking?
Which learner decisions are becoming independent?
Which supports keep returning?
Which tasks are still artificially easy because the environment gives away the route?
Which failures now justify stronger support?

A healthy learning route should show changing support, not simply changing worksheets.

93. The Term Dose Review

At a longer interval, inspect the whole learner system.

Does the learner start tasks with less prompting?
Can the learner choose methods with fewer labels?
Can the learner plan revision with less adult control?
Can the learner identify and repair errors?
Can the learner transfer across contexts?
Can the learner perform when the tutor is absent?

If the answer is yes, the tutor should become less operationally central even while remaining intellectually valuable.

94. Research Foundations

The exact Dose model in this chapter is an eduKate operating framework rather than a single published protocol. Its logic aligns with several established research traditions.

Koedinger & Aleven — Exploring the Assistance Dilemma in Experiments with Cognitive Tutors: assistance can be mistimed or mismatched; the instructional problem is deciding when and how much help improves learning.
van de Pol, Volman & Beishuizen — Scaffolding in Teacher–Student Interaction: A Decade of Research: effective scaffolding is commonly framed around contingency, fading and transfer of responsibility.
Kalyuga, Ayres, Chandler & Sweller — The Expertise Reversal Effect: instructional supports that help less experienced learners can become redundant or counterproductive as expertise increases.
Renkl, Atkinson, Maier & Staley — From Example Study to Problem Solving: Smooth Transitions Help Learning: fading worked solution steps offers a way to move gradually from example study toward independent problem solving.
Soderstrom & Bjork — Learning Versus Performance: An Integrative Review: good immediate performance is not identical to durable learning, so support should be judged by what survives beyond the supported moment.

Together these traditions support a practical conclusion: instructional support should be adaptive, should change with expertise, and should eventually return responsibility to the learner.

95. What the Research Does Not Justify

Research does not justify a universal rule that less guidance is always better.

Nor does it justify a universal rule that more explicit teaching is always better.

Learner knowledge, task complexity, time pressure, prior experience, misconception, transfer demand and performance conditions all matter.

The Dose is therefore a calibration discipline, not an ideology.

96. The Dose State Model

Observe → protect the learner’s cognitive job → choose the smallest useful support → wait for response → measure what changed → increase support only if productive thinking cannot continue → reduce support as ownership rises → retest fresh → change the surface → return after delay → test under realistic load → stop when the capability survives without the scaffold.

The model is cyclical because learner state changes.

97. The Dose Operating Cycle

Name the target mechanism → define the learner job that must remain protected → estimate current expertise → choose the initial dose → define the rescue threshold → apply support → pause → observe whether the learner can resume → escalate only one useful step if needed → require a fresh attempt → fade the support → test changed surface → test delayed return → test authentic conditions → record the lowest dose that still works → retire support when independent control is stable.

98. The Tutor’s Four Dose Questions

Before every meaningful intervention, ask:

What must the learner still do?
What is the least help that makes that work possible?
How will I know this help is now too much?
What will success look like after I remove it?

Those four questions turn help into a designed intervention rather than a reflex.

99. The Final Tutor Standard

A weak tutor proves usefulness by doing more.

A better tutor helps accurately.

A stronger tutor notices how much help is enough.

The strongest tutor notices when yesterday’s useful help has become today’s unnecessary support.

That tutor can wait without neglecting, explain without dominating, guide without steering every move, practise without burying the learner in volume, and withdraw support without making independence collapse.

The measure is not how indispensable the tutor becomes.

The measure is whether the learner can do more of the right work when the tutor does less of it.

Evidence and Connected Reading

Final Compression

Choose the right branch.

Then calibrate the dose.

Give enough help to restore useful thinking, but not so much that the support performs the thinking.

Protect the learner’s cognitive job.

Wait before rescuing when the pause may contain useful work.

Escalate when the learner cannot continue productively.

Fade when the learner can carry more.

Retest without help.

Change the surface.

Return after delay.

Test under the conditions that actually matter.

If success survives, reduce support again.

If success collapses, do not automatically restore everything. Find the lightest support that restarts the learner’s own process.

The tutor is not trying to make every lesson look effortless.

The tutor is trying to make capability increasingly portable.

Good help solves the present problem. Great help changes what the learner can do the next time help is absent.

That is The Dose.

That is Tutor Handbook Volume 0038.

Next in the series: The Tutor Handbook Vol No.0039 | The Fade — How a Tutor Removes Support Without Making Independent Performance Collapse.

Small Group Tutorials

The Tutor Handbook Vol No.0038 | The Dose — How a Tutor Calibrates How Much Help to Give Before Support Starts Replacing Learning