Tutor Handbook Vol 0010 | The Second Lesson

The Tutor Handbook · Volume 0010 · Series ID THB-0010

The first lesson gives the tutor a story.

The second lesson has to decide whether the story deserves to survive.

This is one of the most important transitions in tutoring because first impressions are seductive. A tutor meets a student, sees an error pattern, forms a plausible explanation and immediately wants to build a plan around it.

Perhaps Alicia appears weak at word problems.

Perhaps Beatrice seems unable to infer meaning from a passage.

Perhaps Denise makes repeated algebraic errors.

Perhaps Faith knows her Science content but writes thin explanations.

All four observations may be correct.

The danger begins when the tutor turns them too quickly into permanent identities.

The second lesson should not ask, “How do I continue my plan?” It should first ask, “What evidence would make me change my mind?”

This is Volume 0010 of The Tutor Handbook, eduKate Sengkang’s practical long-form series on the jobs that exist inside tutoring. Volume 0009, The First Lesson, argued that the tutor should leave the opening session with a working hypothesis rather than a permanent diagnosis. Volume 0010 owns the next job:

How should a tutor use the second lesson to test the first explanation, distinguish a real pattern from a one-off performance, and decide whether the learner is ready for a longer route?

The wider tutor classification remains owned by the Tutor Classification Model by eduKateSG and the Tutorial Class Control Tower. The ongoing diagnostic mechanism remains owned by How Learning Diagnosis Works and How Diagnostic Teaching Works. This handbook volume sits between them: it is the transition from first-contact evidence to a defensible tutoring route.

Quick Read

The first lesson should produce a working hypothesis. The second lesson should pressure-test it.
Do not simply repeat the same task and celebrate the same result. Change one meaningful condition.
Useful second-lesson checks include delayed retrieval, changed surface forms, reduced prompting, alternative representations, mixed questions and reattempts of previous errors.
A hypothesis becomes stronger when the same failure signature appears across more than one task or condition and weaker when the predicted pattern disappears.
The tutor should actively look for disconfirming evidence, not only evidence that supports the first impression.
A second lesson can confirm, narrow, revise or reject the first hypothesis.
The tutor should avoid building a long-term plan from one unusually good or unusually bad lesson.
Parent and student reports are still useful, but direct learner performance remains central.
The second lesson should identify which tutor class and tuition mode still fit after new evidence arrives.
The learner should leave with a clearer next step, while the tutor leaves with less uncertainty rather than false certainty.

1. The Second Lesson Is Where Professional Humility Becomes Useful

A weak tutor wants to be right.

A strong tutor wants the explanation to survive evidence.

The difference is subtle but consequential.

During the first lesson, the tutor might conclude:

Alicia’s main difficulty seems to be translating verbal relationships into mathematical representations.

That is useful because it is specific and testable.

It becomes dangerous only when Lesson 2 is designed solely to prove it.

The stronger stance is:

If this is really the bottleneck, what should I observe when the question changes—and what observation would suggest that I am wrong?

The second lesson therefore begins with intellectual humility, not because the tutor knows nothing, but because one session rarely reveals enough to justify a durable learner model.

2. The First Hypothesis Is a Prediction

A useful hypothesis does more than describe yesterday.

It predicts tomorrow.

If Denise’s algebraic errors are caused mainly by weak sign handling, then similar sign losses should appear in another topic that uses the same underlying operation.

If Beatrice’s comprehension weakness is mainly question-demand interpretation, then her answer quality should improve when she correctly paraphrases the demand before reading for evidence.

If Faith’s Science explanations fail because the mechanism link is missing, then asking her to explicitly connect condition → concept → mechanism → outcome should improve changed questions too, not only the exact example taught last week.

The second lesson is where those predictions are tested.

3. Confirmation Is Not Repetition

Suppose Alicia struggled with a fraction word problem in Lesson 1.

In Lesson 2, the tutor gives another nearly identical fraction word problem.

Alicia succeeds.

Has the hypothesis been confirmed?

Not strongly.

The learner may remember the previous format, the tutor’s language, the worked representation or the local pattern of the worksheet.

A better test changes something meaningful while keeping the underlying relationship stable.

The second lesson should test the mechanism, not the memory of the first lesson.

4. Change One Important Condition at a Time

If the tutor changes everything at once, the result becomes hard to interpret.

A useful second lesson can vary one of these:

surface context;
representation;
question wording;
support level;
time delay;
topic mixing;
sequence position;
amount of irrelevant information;
degree of prompting;
requirement to explain rather than merely answer.

Then observe what changes.

This is not laboratory experimentation in a strict scientific sense. Tutoring is too complex and local for that claim.

But the discipline is useful:

If you want to learn what caused the performance, do not alter ten conditions and then guess which one mattered.

5. Start Lesson 2 With a Delayed Return

One of the best openings for the second lesson is a small delayed return to something repaired in the first lesson.

Do not re-explain first.

Do not show the old worked example first.

Ask the learner to reconstruct.

This reveals whether the first lesson changed immediate performance only or whether some useful capability survived the interval.

The result can be classified roughly:

Return state	What it suggests
Returns independently	The repair may be stabilising
Returns after one small cue	Capability is partly available but not yet robust
Returns only after the original explanation is recreated	The first lesson may have produced supported performance rather than durable change
Fails differently	The original diagnosis may have been incomplete

6. Delayed Retrieval Is More Informative Than Immediate Fluency

At the end of Lesson 1, the student may look transformed.

The method is fresh.

The tutor’s language is still active.

The examples are similar.

Lesson 2 removes some of that freshness.

That makes the second lesson unusually valuable.

Retrieval research broadly supports the value of requiring learners to bring knowledge back rather than only restudying it, but the tutor should not overstate what one successful retrieval proves. It is one receipt among several.

7. Reduce the Prompt Before Increasing the Difficulty

Tutors often test progress by giving harder questions.

Sometimes the better test is the same level with less help.

If the tutor had to ask three questions to help Alicia start in Lesson 1, Lesson 2 might begin by asking only one.

If Faith needed the full causal chain written beside her, the tutor might now ask only:

What connects the concept to the outcome?

If Denise needed the tutor to point at every sign transition, Lesson 2 can remove that scaffold and ask her to run her own sign check.

Support reduction tests ownership.

8. The Second Lesson Should Ask Whether the Error Signature Reappears

A single error can be noise.

A recurring error across different tasks begins to look like a pattern.

The tutor should therefore name the predicted signature before the lesson.

For example:

Denise loses negative signs mainly after expansion or transposition.
Beatrice identifies evidence correctly but answers the wrong logical relationship.
Alicia calculates accurately after the representation is chosen but cannot choose the representation independently.
Faith states the concept but omits the mechanism.

Lesson 2 then asks whether that signature appears again under a changed condition.

9. Look for the Predicted Failure—and the Predicted Success

A good hypothesis predicts both.

If Alicia’s problem is representation rather than arithmetic, she should often calculate correctly once the relationship is represented.

If Beatrice’s issue is question interpretation rather than passage comprehension, she should improve when the demand is correctly paraphrased.

If those predicted strengths do not appear, the explanation may be too narrow.

This prevents diagnosis from becoming a search only for weakness.

Strength patterns help locate the boundary of the problem.

10. What Would Falsify the First Hypothesis?

This is one of the most useful questions a tutor can ask privately before Lesson 2.

If the hypothesis is:

The student’s main problem is slow retrieval.

Then evidence against it might be:

the student retrieves quickly but then chooses the wrong method;
the student performs equally slowly even with the needed fact supplied;
the student’s delay appears only on unfamiliar wording, not on direct questions;
the student is fast and accurate untimed but deteriorates only under performance conditions.

The tutor should actively look for such contradictions.

11. Confirmation Bias Is a Tutoring Risk

Once a tutor believes a learner is “weak at algebra”, every algebra mistake starts to look like confirmation.

But a tutor may be seeing only what the label makes visible.

The learner might actually be weak at symbolic reading, sign handling, working-memory organisation, question interpretation or method selection.

The second lesson should therefore include at least one task that could plausibly contradict the first story.

Do not only ask, “Can I find the weakness again?” Ask, “Can I create a fair chance for the weakness not to appear?”

12. The EEF’s 2026 Check–Adapt Principle Fits Lesson 2 Well

The Education Endowment Foundation’s February 2026 Check. Adapt. resource makes a useful point: checking understanding matters because the evidence changes what happens next. The purpose is not the checking routine itself.

That principle maps naturally onto the second tutoring lesson.

The tutor should not run a diagnostic exercise because “Lesson 2 is the diagnostic lesson”.

The tutor should gather evidence because a real decision remains unresolved.

Do we continue?

Do we simplify?

Do we increase support?

Do we fade support?

Do we change the diagnosis?

The value is in the adaptation.

13. Lesson 2 Should Compare Like With Like Before Comparing Hard With Easy

Suppose the first lesson used an easy task and the second lesson uses a difficult paper.

If performance deteriorates, the tutor has learned something—but not necessarily whether the original weakness persisted.

A better sequence often includes:

one comparable item to test return;
one changed item to test transfer;
one harder item to locate the new boundary.

This gives the tutor three levels of evidence instead of one dramatic jump.

14. Reattempt One Old Error Without Showing the Correction

One of the highest-value second-lesson moves is to return to an error from Lesson 1.

Hide the correction.

Ask the learner to reconstruct.

The tutor can then distinguish several possibilities:

The learner remembers the corrected answer only.
The learner remembers the mechanism and can rebuild the answer.
The learner can apply the repair to a changed item.
The learner has forgotten the repair entirely.
The learner now fails somewhere earlier than before.

Each outcome changes what the tutor should do next.

15. The Second Lesson Should Test Feedback Closure

Feedback is incomplete until it changes later performance.

The EEF’s current feedback guidance emphasises that feedback should be grounded in high-quality teaching and evidence about learning, and that learners need opportunities to act on it.

Lesson 2 provides the first natural return point.

If the first lesson’s correction was useful, something should be different now.

Maybe not perfect.

But different.

16. Do Not Measure Only Correctness

Lesson 2 can improve even when the mark stays the same.

Look for:

fewer prompts;
faster independent starts;
better explanation of the first step;
more accurate self-checking;
less repeated error;
better recovery after being stuck;
greater willingness to attempt before asking;
more accurate prediction of difficulty.

These are not substitutes for eventual academic performance.

They are leading indicators of a changing learning process.

17. Compare the Support Conditions Explicitly

A tutor can easily forget how much help was supplied in Lesson 1.

Then Lesson 2 looks worse simply because support has been reduced.

Record enough to compare fairly.

Condition	Lesson 1	Lesson 2
Prompt level	Three cues	One cue
Question familiarity	Worked example nearby	No model visible
Timing	Untimed	Untimed
Surface form	Familiar	Changed
Outcome	Correct	Correct after self-correction

Lesson 2 may actually show stronger learning despite looking messier.

18. Messier Independent Work Can Be Better Evidence Than Smooth Supported Work

This is one of the most important things parents and tutors need to understand.

A student who gets eight out of ten correct independently may be in a stronger state than a student who gets ten out of ten with constant prompts.

The second lesson should therefore tolerate some temporary visible imperfection as support fades.

Do not rush to restore the polished page.

First ask what the learner is now carrying alone.

19. The Second Lesson Should Re-read the Learner State

A learner can arrive in a different state from the first meeting.

The first lesson may have been after a long school day.

The second may happen on a weekend.

The learner may be less nervous.

Or more tired.

The tutor should notice these contextual differences before over-interpreting performance changes.

A learning hypothesis should be robust enough to survive ordinary state variation—but the tutor should not pretend conditions are identical when they are not.

20. One Good Lesson Does Not Erase the Original Problem

Lesson 2 sometimes goes beautifully.

The student remembers everything.

The parent feels relieved.

The tutor is tempted to conclude that the first concern was exaggerated.

Be careful.

Performance can improve because the learner now understands the tutor’s expectations, because the topic is familiar, because the second session occurs under better conditions or because the first lesson genuinely worked.

The correct response is not disbelief.

It is continued testing.

21. One Bad Lesson Does Not Prove the Plan Failed

The reverse is also true.

The learner may have forgotten more than expected.

The new representation may be genuinely harder.

The learner may have had a difficult school day.

The tutor may have reduced support too aggressively.

The hypothesis may still be right while the intervention needs adjustment.

Lesson 2 should update the model, not trigger panic.

22. The Second Lesson Has Four Possible Outcomes

Outcome	Meaning
Confirm	The predicted failure and strength patterns reappear under changed conditions
Narrow	The original hypothesis was broadly right but too general
Revise	New evidence points to a different upstream cause
Reject	The predicted pattern does not survive fair retesting

All four can be successful second lessons.

Rejecting a wrong hypothesis is progress.

23. Confirm: When the Pattern Reappears Cleanly

Alicia again solves the calculations correctly once the relationship is represented, but struggles to construct the representation herself in two different word-problem contexts.

The first hypothesis becomes stronger.

The tutor can now design a route around relationship interpretation and representation rather than generic arithmetic remediation.

24. Narrow: When the Hypothesis Was Right but Too Broad

Beatrice appears weak at inference in Lesson 1.

In Lesson 2, she handles attitude and purpose well but repeatedly fails cause-and-effect inference.

“Weak at inference” is now too broad.

The better description is narrower and more teachable.

25. Revise: When the New Evidence Moves the Weak Link Upstream

Denise appears weak at trigonometry because many trigonometry questions are wrong.

Lesson 2 reveals that she identifies trigonometric relationships correctly but makes the same algebraic rearrangement errors in unrelated equations.

The working diagnosis shifts upstream.

The tutor should not become loyal to “trigonometry weakness” simply because that was the first visible topic.

26. Reject: When the Pattern Does Not Survive

Faith appeared weak at causal explanation in Lesson 1.

In Lesson 2, she explains changed Science problems accurately and independently.

Closer inspection shows the original poor answer occurred after she misread one condition in the question.

The tutor should reject the broader explanation and investigate question reading instead.

This is good tutoring.

The tutor learned that the first story was too large.

27. Lesson 2 Should Reclassify the Tutor Job If Necessary

The first lesson may suggest Class 1 explanation.

The second may reveal Class 3 diagnosis.

Or the reverse.

A student who appeared confused may understand perfectly once a missing vocabulary term is clarified. The dominant problem may now be practice.

A student who appeared merely under-practised may repeat the same misconception despite extensive work. Diagnosis becomes necessary.

Use the tutor ladder as a live classification, not a permanent label.

28. Lesson 2 Should Reclassify the Tuition Mode If Necessary

The first lesson may suggest Repair.

Lesson 2 may show that the foundation is stronger than expected and the real problem is Alignment with current school work.

Or a high-scoring student may appear ready for Frontier work until changed questions reveal fragile transfer.

The tutor should be willing to move the learner between Repair, Alignment and Frontier based on evidence.

29. The Second Lesson Should Not Become the Long-Term Plan Yet

Two sessions are better than one.

They are still only two sessions.

The tutor can begin constructing a short route—perhaps two to six weeks—without pretending the next year is fully understood.

A strong plan at this stage is deliberately revisable.

Commit to the next useful route, not to a permanent story about the learner.

30. Build the Smallest Route That Can Test the Hypothesis

If Alicia’s bottleneck appears to be representation, do not design a six-month Mathematics programme immediately.

Build a short cycle:

relationship reading → representation → method selection → independent attempt → changed problem → delayed return.

If the hypothesis is correct, performance should change in predictable ways.

If not, the short route will reveal the mismatch before months are wasted.

31. Lesson 2 Should Preserve Current School Functionality

Diagnosis can become so interesting that the tutor forgets the learner still has school tomorrow.

Lesson 2 should check current assignments, upcoming assessments and the active school topic.

Then balance:

the hypothesis test;
the first repair;
the current school obligation.

The EEF’s one-to-one tuition evidence continues to emphasise connection with normal classroom teaching. This does not mean tuition must mirror school exactly. It means the learner’s real curriculum should remain visible in tutoring decisions.

32. Do Not Let Current Homework Hijack the Entire Second Lesson

The opposite failure also occurs.

The student arrives with urgent homework.

The tutor spends the whole session finishing it.

The first hypothesis is never retested.

The tutor returns to episodic support.

A good second lesson can use the homework itself as evidence where possible.

If not, reserve enough time for one deliberate hypothesis test.

33. Alicia: The Word-Problem Hypothesis

Lesson 1 suggests Alicia can calculate fractions but cannot reliably translate verbal relationships into a representation.

Lesson 2 begins with a delayed changed problem.

Alicia again hesitates before representing.

The tutor does not say, “Use a model.”

Instead:

What is the whole, and what changed?

Alicia recovers.

Then she receives a different context with the same relationship and starts independently.

The hypothesis survives but changes resolution.

Alicia does not lack all representation skill.

She needs a cue when the relationship is verbally dense.

That is a better teaching target.

34. Beatrice: The Inference Hypothesis

Lesson 1 suggests Beatrice struggles with inference.

Lesson 2 deliberately separates three jobs:

literal retrieval;
cause-and-effect inference;
attitude inference.

Beatrice performs well on literal retrieval and attitude but repeatedly answers cause questions with sequence descriptions.

The first hypothesis narrows.

The route now trains causal relationships rather than “inference” as one giant category.

35. Denise: The Trigonometry Hypothesis

Lesson 1 appears to show weak trigonometry.

Lesson 2 gives Denise one trigonometry question and one unrelated equation requiring the same algebraic rearrangement.

The same sign error appears in both.

The tutor revises the hypothesis.

The route moves upstream to algebraic control before returning to timed trigonometry.

36. Faith: The Science-Explanation Hypothesis

Lesson 1 suggests Faith knows keywords but omits mechanism.

Lesson 2 begins with a different Science context.

Faith states the condition, concept and mechanism correctly without a model answer.

Then another question contains a misleading diagram label. Faith misreads the label and gives a wrong explanation.

The broad mechanism hypothesis weakens.

The tutor now investigates condition reading and evidence extraction.

Lesson 2 has prevented weeks of unnecessary model-answer practice.

37. Quietly Weak Learners Need More Unsupported Moments in Lesson 2

A quietly weak learner may have looked better in Lesson 1 because the tutor was unusually attentive.

Lesson 2 should include periods where the tutor deliberately attends elsewhere or remains silent.

Observe whether the learner continues, waits, copies, restarts, checks or asks.

This is especially useful in a three-pax class, where independent continuation is naturally required.

38. Hardworking but Stuck Learners Need Mechanism Tests, Not More Volume

If the learner already completed twenty practice questions after Lesson 1 and the same failure returns, Lesson 2 should not prescribe twenty more immediately.

Ask:

Was the practice at the correct layer?
Did the learner receive feedback?
Was the correction reattempted?
Did the task vary enough to test transfer?
Was the learner practising an error repeatedly?

Lesson 2 should discover why practice failed to convert before adding more practice.

39. “Careless” Learners Need a Condition Test

If Lesson 1 suggests carelessness, Lesson 2 should change one condition.

Remove time pressure.
Use fewer questions.
Require a targeted sign or unit check.
Compare early and late-session performance.
Ask the learner to verbalise the first decision.

If the errors disappear, “careless” was too broad.

If they persist under calm conditions, the tutor has different evidence.

40. Strong Learners Need a Boundary Test

A strong first lesson can tempt the tutor to accelerate immediately.

Lesson 2 should test depth before speed.

Can the learner justify?
Can they compare methods?
Can they solve after representation changes?
Can they identify a flawed solution?
Can they transfer without the topic label?

If yes, Frontier work may be justified.

If not, the student may be high-performing inside familiar conditions rather than genuinely transfer-ready.

41. The Second Lesson Should Check Whether the Learner Used the Home Task

The first lesson often ends with a small home task.

Lesson 2 should inspect not only whether it was completed but how.

Did the learner attempt before looking at help?
Where did they stop?
What did they do when stuck?
Did they use the tutor’s new rule?
Did they self-correct?
Did a parent provide substantial help?
Did an answer key or AI tool change the support condition?

The home task is part of the evidence chain.

42. Completion Without Process Evidence Is Weak Evidence

The homework is complete.

That tells the tutor something.

It does not tell the tutor who did the thinking.

Lesson 2 should reconstruct enough of the process to know whether the home performance supports or weakens the first hypothesis.

43. The Second Lesson Should Ask the Student What Felt Different

Students may notice changes the tutor cannot see.

“I knew how to start this time.”
“I still understood it at home, but I forgot the second step.”
“The new question looked different and I panicked.”
“I could do it until I had to explain why.”
“I did not need to ask my mother.”

These reports are not objective measurements.

They are useful clues to the learner’s developing self-model.

44. Metacognitive Talk Should Stay Inside the Subject

The EEF’s 2026 work on metacognitive talk emphasises helping learners articulate how they approach tasks, monitor progress and evaluate strategies within ordinary subject teaching.

Lesson 2 can use this productively.

Instead of asking a generic “How do you feel about learning?”, ask:

Why did you choose this method?
What told you the first approach was failing?
What did you check before moving on?
Which part of last lesson did you remember independently?
What will you try if this happens again?

The conversation remains attached to real academic decisions.

45. The Student Should Begin Predicting Too

Before a changed task, ask the learner:

Which part do you think will be hardest, and why?

Then compare the prediction with what happens.

This helps the learner develop a more accurate model of their own strengths and weaknesses.

The EEF’s August 2026 Developing Independent Learners materials describe planning, monitoring and evaluating learning as processes that can gradually become internal scaffolding. Lesson 2 is an early place to begin transferring some of the diagnostic language to the learner.

46. In a Three-Pax Class, Lesson 2 Can Test Independence Naturally

Small-group tuition offers an important second-lesson advantage.

The tutor can deliberately stop being continuously available.

While the tutor works with Beatrice, Alicia must continue.

While Alicia receives a short explanation, Faith must check her own answer.

This creates a natural test:

Does the learner know what to do next when tutor attention is elsewhere?

The National Student Support Accelerator’s high-impact tutoring design materials continue to emphasise small groups, consistent tutor relationships, progress monitoring and curricular alignment as useful design features. The local implementation question remains whether those features are producing the intended learner behaviour.

47. One-to-One Lesson 2 Needs Deliberate Silence

One-to-one tutoring creates extraordinary visibility.

It also makes over-support easy.

In Lesson 2, the tutor may need to deliberately create independent intervals.

Set the task.

Say nothing.

Observe.

Then intervene only when the evidence requires it.

Otherwise Lesson 2 may measure the tutor–student pair rather than the learner.

48. The Second Lesson Should Clarify the Tutor–Student Fit

Volume 0008 argued that fit is dynamic.

Lesson 2 provides the first real chance to test it.

Does the learner reveal uncertainty to this tutor?
Does the tutor wait long enough to observe?
Is the support level appropriate?
Can the tutor explain at the learner’s current resolution?
Does the learner become more independent or more dependent?
Does the class structure make the relevant evidence visible?

One lesson establishes rapport.

Two lessons begin to reveal whether the relationship produces useful learning evidence.

49. The Parent Report After Lesson 2 Should Change

After Lesson 1, the tutor may say:

My current hypothesis is…

After Lesson 2, the report can become more precise:

The same pattern appeared in two changed tasks and reduced after one specific cue, so I am more confident that this is the first useful weak link. I still want one delayed return before expanding the route.

Or:

The first hypothesis did not hold. The learner’s concept knowledge is stronger than it looked initially; the main problem now appears to be interpreting the condition in the question.

Both reports demonstrate progress.

50. Parents Should Hear What Changed the Tutor’s Mind

A trustworthy tutor can explain not only the conclusion but the evidence that altered it.

This makes tutoring legible.

It also protects families from vague narratives that can explain every outcome after the fact.

51. The Student Should Hear What Improved

Do not only report deficits.

Say what changed in observable terms.

“Last week you needed me to identify the relationship. Today you did it yourself on two questions.”
“You still lost one sign, but you caught it during your own check.”
“You remembered the Science mechanism after a week without seeing the model answer.”
“You were able to explain why your first English answer did not match the question.”

Specific progress helps the learner build a more accurate self-model.

52. The Second Lesson Should End With a Short Route, Not a Giant Programme

Once the hypothesis is stronger, design the next few lessons around one dominant transformation.

For example:

From “needs a prompt to identify the mathematical relationship” to “can represent and select a method independently across changed problems”.

Or:

From “copies the corrected Science mechanism” to “can reconstruct the causal chain after a delay and apply it in an unfamiliar context”.

A short route is easier to evaluate honestly.

53. The Second-Lesson Homework Should Test the Hypothesis Too

Do not assign generic extra work simply because tutoring traditionally ends with homework.

A useful home task can test one thing:

Can the learner start without the tutor?
Can the learner use the new checking rule?
Can the learner retrieve after another delay?
Can the learner apply the repair to a different surface form?
Can the learner record exactly where help becomes necessary?

Homework becomes another observation point rather than mere volume.

54. Record the Smallest Useful Prompt

One of the most valuable second-lesson records is not the score.

It is the smallest prompt that restored productive movement.

This tells the next lesson where to begin support.

If one short cue works, the tutor should not return automatically to a full explanation next week.

Support should match evidence.

55. Record the Strongest Independent Evidence

Also record the best thing the learner did alone.

This matters because tutors can become deficit collectors.

The learner model needs stable strengths as well as weak links.

A route becomes more efficient when it knows what does not need repairing.

56. Record What Is Still Unknown

After two lessons, uncertainty remains.

Write it down.

Does the improvement survive another week?
Does the error reappear only under timing?
Does the learner transfer across subjects or only within one topic?
Is the support dependence specific to this tutor?
Will current school demands destabilise the repair?

Unknowns are not defects in the tutor.

They are part of an honest learner model.

57. Failure Mode: Repeating Lesson 1 More Efficiently

The tutor believes the first diagnosis and simply teaches the same material better.

The learner may improve.

The tutor learns almost nothing about whether the diagnosis was right.

Lesson 2 should include at least one deliberate test of the first story.

58. Failure Mode: Making the Second Lesson Harder Just to Test the Student

Harder is not always more informative.

If every variable changes, failure becomes ambiguous.

Use controlled increases in difficulty where possible.

59. Failure Mode: Searching Only for Confirmation

The tutor sees what they expect to see.

Every hesitation becomes evidence of the chosen weakness.

Every success is ignored as a fluke.

This is how a working hypothesis becomes a fixed label.

Actively design one task that could prove the explanation too broad or wrong.

60. Failure Mode: Changing the Hypothesis After Every Single Item

The opposite error is hyper-reactivity.

One correct answer and the weakness disappears.

One new error and the diagnosis changes again.

Look for patterns across several observations.

A learner model should update.

It should not oscillate wildly.

61. Failure Mode: Forgetting the Support Condition

“She got it right last week.”

Yes—with a worked example open and the tutor asking four questions.

This week she is working independently.

The comparison must include support.

62. Failure Mode: Treating a Better Relationship as Better Learning

Lesson 2 often feels smoother because the learner knows the tutor.

That is valuable.

It may improve honesty, willingness to attempt and tolerance of correction.

Still ask what changed academically.

Relationship is part of the condition, not the whole receipt.

63. Failure Mode: Letting the Parent Story Dominate New Evidence

If the parent says the child is careless, the tutor may interpret every second-lesson error through carelessness even when the pattern points elsewhere.

Parent observations remain useful.

Lesson 2 should refine them, not merely reproduce them.

64. Failure Mode: Turning the Second Lesson Into a Sales Proof

The tutor may feel pressure to show immediate transformation.

That pressure can produce over-help.

The page looks perfect.

The parent is impressed.

The learner’s independent state remains invisible.

A stronger second lesson demonstrates value through better resolution and a more defensible next step.

65. Failure Mode: Building a Six-Month Programme From Two Data Points

A long plan can look professional.

At this stage it may be false precision.

Build the next small route.

Set a review point.

Let later evidence earn later detail.

66. A Practical Second-Lesson Architecture

There is no universal schedule, but a useful 90-minute architecture might be:

Phase	Job
Delayed return	Test what survived from Lesson 1 without re-teaching first
Changed condition	Alter one meaningful variable and watch whether the predicted signature returns
Disconfirmation check	Give the hypothesis a fair chance to be wrong
Targeted teaching	Repair the now-better-defined weak link
Reduced-support reattempt	Check whether the learner can carry more of the process
Current-school alignment	Connect the finding to real homework or school demands
Close	Confirm, narrow, revise or reject the hypothesis and set the next test

The exact minutes depend on age, subject and urgency.

The architecture matters more than rigid timing.

67. The Second-Lesson Evidence Card

Field	What to record
Lesson 1 hypothesis	The exact predicted bottleneck
Return result	What survived after the delay?
Changed condition	What variable was deliberately altered?
Predicted signature	What failure or strength was expected?
Observed signature	What actually appeared?
Smallest useful prompt	What restored progress?
Disconfirming evidence	What did not fit the first story?
Hypothesis status	Confirm, narrow, revise or reject
Current tutor class	What tutoring function now dominates?
Next test	What remains unresolved?

This record is intentionally small.

The purpose is continuity, not bureaucracy.

68. What the Third Lesson Should Inherit

Lesson 3 should inherit more than a worksheet.

the updated hypothesis;
the strongest independent evidence;
the recurring failure signature;
the smallest useful prompt;
the current short route;
the uncertainty still alive;
the next condition to test.

This is how tutoring becomes cumulative.

Each lesson should inherit the unresolved question from the previous one.

69. When Lesson 2 Says “Stop and Diagnose More”

Sometimes the second lesson makes the picture less clear.

Different failure signatures appear.

Performance changes unpredictably across representations.

Support helps one task and disrupts another.

This is not a reason to force a route.

Return to Class 3 diagnostic work.

The best next move may be more discrimination, not more teaching.

70. When Lesson 2 Says “Start Building the Route”

If the pattern repeats, the intervention helps, the changed task improves and the support boundary is becoming clear, Class 4 can begin.

The Route Designer can now sequence:

repair → guided practice → independent practice → delayed return → mixed use → performance.

The route is not yet permanent.

It is now evidence-based enough to deserve a trial.

71. When Lesson 2 Says “The Learner Needs Less Help Than We Thought”

This is an important success state.

The first lesson may have overestimated the need because the student was nervous, unfamiliar with the tutor or facing an atypical task.

Lesson 2 may show that the learner is substantially independent.

Do not invent need to justify tutoring.

Shift toward Alignment, Frontier or a lighter-touch arrangement if that is what the evidence supports.

72. When Lesson 2 Says “The Learner Needs More Support Than We Thought”

The opposite may happen.

The first lesson may have sampled familiar content and hidden a deeper break.

Changed tasks reveal that the student cannot transfer, cannot retrieve after delay or depends heavily on prompts.

Do not punish the learner with harder work.

Increase diagnostic resolution and select the correct tutor class.

73. The Tutor Should Explain Why the Plan Changed

Students can interpret a changed plan as inconsistency.

Parents can interpret it as uncertainty.

Explain the evidence.

Last lesson I thought the main problem was topic knowledge. Today you retrieved the knowledge accurately but struggled when the representation changed. So I am changing the next step.

That is not weakness.

That is responsive teaching made visible.

74. Twelve Questions Every Tutor Should Answer After Lesson 2

What was my first hypothesis?
What did I predict would reappear?
What condition did I change?
What survived after the delay?
What support was still necessary?
What evidence supported the hypothesis?
What evidence contradicted it?
Did the hypothesis confirm, narrow, revise or fail?
What tutor class now owns the dominant job?
What short route should we test next?
What remains uncertain?
What evidence would make me change my mind again?

75. Twelve Questions Parents Can Ask After Lesson 2

Did the same problem appear again?
What changed from the first lesson?
What did my child remember independently?
What support was still needed?
Did anything contradict the first diagnosis?
Is the original explanation now stronger or weaker?
What are you working on next?
Why is that the next step?
What should my child do before the next lesson?
What should we avoid helping with at home?
When will this short route be reviewed?
What would make you change the plan?

76. Frequently Asked Questions

Why is the second lesson different from the first?

The first lesson establishes initial evidence and a working hypothesis. The second lesson tests whether that hypothesis survives delay, changed conditions and reduced support. It is the first chance to distinguish a recurring mechanism from a one-off first-session performance.

Should the tutor repeat the same material?

Some comparable material is useful for checking return, but the second lesson should also change at least one meaningful feature so the tutor can test transfer rather than simple repetition.

What if the student forgets everything from Lesson 1?

That is important evidence. Determine whether a small cue restores access, whether the learner recognises the earlier method, and whether the original teaching reached understanding or mainly produced immediate supported performance.

What if the student performs much better?

That is encouraging, but do not over-interpret one strong session. Compare support conditions and test a changed task before concluding the original weakness has disappeared.

What if the student performs worse?

Look at what changed: difficulty, representation, timing, fatigue, support or topic. A worse score may reveal fragility, but it does not automatically prove the original plan failed.

When can the tutor begin a longer-term plan?

Begin with a short revisable route once the main failure signature appears consistently enough to justify targeted work. Longer plans should gain detail as repeated evidence accumulates.

Is changing the diagnosis a bad sign?

No. Changing a working hypothesis because new evidence contradicts it is a sign of responsive professional judgement. The problem is not changing the hypothesis; it is refusing to change when the evidence no longer fits.

How much homework should follow Lesson 2?

Enough to test or strengthen the current hypothesis. A small high-information task can be more useful than a large generic worksheet.

Should parents help with the second-lesson homework?

Follow the tutor’s support instructions. If the purpose is to test independent performance, heavy parent help can destroy the evidence. If the learner becomes genuinely stuck, record where help was needed rather than hiding it.

What should the student understand after Lesson 2?

The learner should have a clearer idea of the current problem, what strategy is being tested, what they should attempt independently and what the tutor will check next.

What is the biggest second-lesson mistake?

Turning the first impression into a fixed diagnosis and using every new task only to confirm it. Lesson 2 should give the first hypothesis a fair chance to be wrong.

What is the clearest sign the second lesson succeeded?

The tutor can make a better next decision because uncertainty has been reduced, the hypothesis has been updated honestly, and the learner has begun carrying more of the relevant process independently.

77. Evidence and Further Reading

78. Final Compression

The first lesson asks:

What might be happening?

The second lesson asks:

Does that explanation still deserve to guide us?

Return after a delay.

Change one condition.

Reduce one prompt.

Look for the predicted error.

Look for the predicted strength.

Give the hypothesis a chance to fail.

Then decide.

Confirm it.

Narrow it.

Revise it.

Reject it.

All four are better than pretending the first impression was certainty.

Then build the smallest useful route that can continue testing the learner while helping them improve.

The second lesson succeeds when the tutor is less attached to the first story and more capable of making the next decision from evidence.

That is the discipline of the second lesson.

That is Tutor Handbook Volume 0010.

Next in the series: The Tutor Handbook Vol No.0011 | The First Month — How Repeated Lessons Turn Early Hypotheses Into a Stable Teaching Route.

Small Group Tutorials

The Tutor Handbook Vol No.0010 | The Second Lesson — Test the First Hypothesis Before Turning It Into a Long-Term Plan