The Tutor Handbook Vol No.0023 | The Second Full Paper

The Tutor Handbook · Volume 0023 · Series ID THB-0023

The first full paper found the problem.

The tutor repaired it.

Now comes the dangerous moment.

The student sits another full paper.

The score goes up.

Everyone wants to celebrate.

Perhaps they should.

But a tutor still has one job left before calling the repair secure:

Did the repaired mechanism actually survive when it was returned to the whole examination system?

This is Volume 0023 of The Tutor Handbook, eduKate Sengkang’s long-form series on the practical decisions inside tutoring.

Volume 0022, The Full Paper, showed how a tutor reads one complete paper as a whole-system trace: score, sequence, timing, method selection, cascades, first weak link and repair queue.

This volume owns the next problem.

After a targeted repair appears successful in small tasks, how does a tutor decide whether it still works inside a fresh, mixed, timed, independent full paper?

What This Volume Owns—and What It Does Not

The wider architecture of examination performance already belongs to How Examination Performance Works. Practice-paper study strategy belongs to How Studying From Practice Papers Works. Diagnosis belongs to How Learning Diagnosis Works. Evidence interpretation belongs to How Assessment Evidence Works. Calibration belongs to How Learning Calibration Works.

This handbook does not replace those pages.

It owns a narrower tutor-operational job: comparing a repaired learner with a later whole-paper performance carefully enough to decide whether the repair survived reintegration, partially survived, changed form, or failed under load.

The tutor classification remains anchored in the Tutor Classification Model by eduKateSG: Class 0 Homework Helper, Class 1 Explainer, Class 2 Drill Builder, Class 3 Diagnostic Tutor, Class 4 Route Designer, Class 5 Performance Coach and Class 6 Learning Architect.

The Second Full Paper is not a new tutor class. It is a verification surface where the existing classes must agree on one discipline:

Do not confuse a better result with proof that the intended repair caused the improvement.

Quick Read

The first full paper discovers a whole-system weakness. The second full paper tests whether the repair survives reintegration.
Define the repair claim before the second paper begins.
Compare the mechanism, not only the total score.
A higher score can occur even when the repaired weakness returns.
A lower score can occur even when the repaired mechanism genuinely improved.
The second paper should be fresh enough to reduce answer-memory effects but comparable enough to make interpretation useful.
Keep timing, supports and tutor intervention reasonably consistent when comparison matters.
One second paper is evidence, not a trend.
Track the target mechanism across opening, middle, late-paper and recovery conditions.
Watch for compensation: one improvement can create a new cost elsewhere.
Watch for substitution: the student may avoid the repaired method rather than master it.
Watch for prompt dependence: a repair that survives only after reminders has not fully reintegrated.
Use fresh small checks when the second paper produces ambiguous evidence.
Distinguish four outcomes: survived, partially survived, displaced, or failed.
The tutor’s job is to decide what the evidence changes next.

1. A Second Full Paper Is Not Merely Another Paper

When a learner completes a first paper, the tutor asks, “What happened?”

When the learner completes a second paper after targeted repair, the question changes.

Did what we changed remain available when everything else returned?

Everything else matters.

Topic labels are gone. The clock is running. Easy questions sit beside difficult ones. Earlier mistakes affect confidence. The learner must decide when to persist, when to move, what to retrieve, which method to select, how much working to show, what to check and how to recover.

A repair that worked beautifully in a five-question lesson may still collapse when returned to this environment.

That is not necessarily bad news.

It is information about the next stage of learning.

2. The Repair Claim Must Be Written Before the Test

Suppose the first paper showed repeated percentage errors.

“Improve percentages” is too vague.

A stronger repair claim is:

When a percentage question changes the reference quantity, the learner will identify the correct base before calculating and will preserve that choice under mixed and timed conditions.

Now the tutor knows what to look for in the second full paper.

The claim has an object, a condition, a decision and an expected behaviour.

Without that clarity, the second paper becomes a fishing expedition.

3. The Second Paper Tests Reintegration

Reintegration means returning a repaired component to the larger system that originally exposed its weakness.

A learner may have repaired equation balance in isolation.

Reintegration asks whether equation balance still works when:

the question is not labelled as an equation exercise;
the numbers are less friendly;
another topic appeared immediately before it;
the learner has already been working for forty minutes;
the clock matters;
the tutor is silent;
the student must choose the method independently;
checking competes with remaining questions.

That is a different achievement from performing the repair in a protected lesson.

4. Repair Locally, Prove Globally

A useful progression is:

targeted repair → fresh targeted check → changed question → mixed set → timed section → second full paper.

Not every repair needs every stage at the same intensity.

A strong learner may move quickly.

A fragile learner may need several mixed returns before a complete paper is informative.

The point is not ritual.

The point is increasing ecological load gradually enough that the tutor can see where the repair stops surviving.

5. A Better Score Is Encouraging, Not Self-Explaining

Paper 1: 61.

Paper 2: 72.

That is worth noticing.

But eleven marks can come from many places.

The repaired weakness may genuinely have improved.
The second paper may have suited the learner better.
More familiar topics may have appeared.
The learner may have guessed more successfully.
Timing may have been more generous.
One difficult question from the first paper may simply be absent.
The learner may have improved in an unrelated area.
The tutor may have given more prompts.

The higher score is good evidence that the second performance was stronger overall.

It is not automatic proof of the repair mechanism.

6. A Lower Score Does Not Automatically Mean the Repair Failed

The reverse mistake is just as common.

Paper 1: 72.

Paper 2: 68.

Suppose the target repair was algebraic sign control.

On Paper 1, six sign errors cost nine marks.

On Paper 2, the learner makes no sign errors but loses marks in a difficult geometry section never targeted by the repair.

The overall score fell.

The repair may still have survived.

The total score tells you how the whole paper went. The target trace tells you what happened to the repair.

7. Compare Like With Like—but Do Not Demand Clones

A perfect comparison would require two papers identical in difficulty, coverage, structure, marking, timing demands and every other meaningful feature while still being completely fresh.

Real school practice rarely gives you that.

So the tutor aims for reasonable comparability, not imaginary perfection.

similar level;
similar examination format;
similar total time;
similar broad topic breadth;
similar calculator or resource rules;
similar marking expectations;
fresh enough that answer memory is not the main driver.

Then interpret cautiously where the papers differ.

8. The Same Paper Reattempt and the Second Full Paper Are Different Instruments

A same-paper reattempt asks:

Can the learner now execute corrected routes on known material?

A fresh second paper asks:

Can the learner select and execute the repair when the answers, order and surface are no longer familiar?

Both can be useful.

They support different claims.

This distinction continues the logic of Volume 0018, The Reattempt, and Volume 0019, The Changed Question.

9. Freshness Protects the Meaning of Success

If a learner remembers that Question 7 used substitution, successful substitution tells you less about independent method selection.

If a learner remembers the model answer’s wording, a polished Science explanation tells you less about reconstruction.

If a learner has seen the comprehension passage repeatedly, improved inference may partly reflect familiarity with the text.

Freshness removes some of these cues.

It does not need to create a completely alien task.

The tutor wants the same underlying job with enough surface change to require genuine selection again.

10. Time Between Papers Changes What the Result Means

A repair demonstrated ten minutes after teaching is different from a repair demonstrated ten days later.

Delay increases the demand on retrieval and retention.

But delay also allows school lessons, homework, revision and other experiences to affect performance.

Record the interval.

Do not treat time as invisible.

A tutor does not need a laboratory schedule. The tutor does need enough awareness to distinguish immediate correction from retained capability.

11. Keep the Comparison Condition Honest

If Paper 1 was independent and Paper 2 involved frequent hints, the comparison changed.

If Paper 1 was completed after a full school day and Paper 2 on a rested weekend morning, the learner state changed.

If Paper 1 had strict timing and Paper 2 received thirty extra minutes, the performance condition changed.

Sometimes those differences are unavoidable.

Record them.

The goal is not to pretend every condition can be standardised.

The goal is to know what you are comparing.

12. Preserve the Independent Run

The second paper is especially valuable when the tutor can resist rescuing the repair.

Suppose the learner reaches the exact kind of question that was repaired.

The tutor notices hesitation.

Do not automatically say:

Remember what we did last week.

That reminder may convert a test of independent retrieval into a test of prompted recognition.

Observe first.

Teach after the run.

13. The Target Trace

Before marking the whole paper, locate every place where the repaired mechanism should have appeared.

Create a target trace.

Target opportunity	What the tutor records
Recognition	Did the learner notice the relevant structure?
Selection	Did the learner choose the repaired rule or method?
Execution	Was it carried out accurately?
Speed	Was the repaired process reasonably efficient?
Independence	Was it done without prompting?
Transfer	Did it work when the surface differed?
Load	Did it survive mixed and timed conditions?
Recovery	Did it return after an earlier difficult item?

Now the repair has a visible path through the paper.

14. A Repair Can Survive in One Form and Fail in Another

The learner may execute the repaired method correctly whenever it is selected but still fail to recognise when to use it.

That is not full survival.

The learner may recognise the method but take so long to reconstruct it that later questions are damaged.

That is not full survival either.

The learner may use the method accurately early in the paper but abandon it under late time pressure.

The repair survived conceptually but not under scarcity.

This is why reintegration is not a yes-or-no checkbox.

15. Four Reintegration Outcomes

Outcome	Meaning	Likely next move
Survived	Repair appears independently in fresh, mixed and timed conditions without meaningful new cost.	Maintain lightly and move to the next priority.
Partially survived	Repair works in some conditions but weakens under novelty, time, switching or late-paper load.	Train the failure condition specifically, then retest.
Displaced	Old error reduced but a compensating or substitute problem appeared.	Diagnose the new cost before declaring success.
Failed	Original mechanism returns materially under representative conditions.	Re-open diagnosis or rebuild the repair route.

16. Survived Means More Than “Got It Right Once”

One correct target question is encouraging.

But suppose the paper contains five opportunities to apply the repaired idea.

The learner gets one right, avoids two, misclassifies one and reaches the last only after a lucky guess.

The tutor should not call the repair secure because one answer was correct.

Look for recurrence across opportunities.

Not statistical perfection.

Enough repeated evidence to support a practical next decision.

17. Partial Survival Is Often the Most Useful Result

Partial survival tells you where the repair breaks.

For example:

works untimed, fails timed;
works when topic is obvious, fails when mixed;
works early, fails late;
works in Mathematics notation, fails in word problems;
works orally, fails in written explanation;
works in familiar contexts, fails when representation changes.

Now the next lesson can target the condition rather than reteach the entire concept.

18. Displacement: When Fixing One Problem Moves the Cost Somewhere Else

A learner is taught to check every algebraic line carefully.

Sign errors fall dramatically.

But the learner now spends so long checking routine work that the final section is incomplete.

The repair is real.

The new system is still not optimal.

This is displacement.

Do not reverse the repair automatically.

Adjust its cost.

A good repair removes the original leak without creating a larger leak elsewhere.

19. Substitution: When the Learner Avoids the Repaired Route

Suppose a student repeatedly chose an unstable algebraic method.

The tutor teaches a cleaner route.

On the second full paper, the student does not make the old mistake.

Good?

Maybe.

Look closer.

The learner skipped every question requiring that decision.

The error disappeared because the task was avoided.

That is substitution, not mastery.

20. Compensation: When a Stronger Habit Hides a Still-Weak Mechanism

A learner may compensate for weak retrieval by spending extra time reconstructing every formula.

The answers become correct.

The total paper slows.

Or the learner may compensate for uncertain comprehension by rereading the passage repeatedly.

Accuracy improves while time collapses.

Compensation can be useful temporarily.

But the tutor should know whether the repaired capability became cheaper or whether the learner simply bought accuracy with another resource.

21. The Second Paper Needs a Before–After Mechanism Map

Mechanism	Paper 1	Repair phase	Paper 2
Reference quantity	Misidentified repeatedly	Explicit base-selection routine	Correct on 4/5 fresh opportunities
Method selection	Often late	Mixed unlabeled practice	Improved early, one late-paper miss
Time cost	Very high	Fluency and decision compression	Reduced materially
Checking	Absent	Targeted high-risk checks	Present but overused

This is far more informative than:

61 → 72.

22. Count Opportunities, Not Just Mistakes

Suppose the target weakness appears twice in Paper 1 and eight times in Paper 2.

Raw error counts are misleading.

Two errors out of two opportunities is different from two errors out of eight.

The tutor does not need sophisticated statistics.

Simple opportunity awareness already improves interpretation.

Ask:

How often did the target mechanism appear?
How often was it recognised?
How often was the repaired action selected?
How often was execution correct?
Under what conditions did failures cluster?

23. Compare Position in the Paper

A repair may look secure because every target opportunity happened in the first thirty minutes.

The tutor still does not know whether it survives late-paper scarcity.

If later papers contain the same decision near the end, watch carefully.

Performance under load is part of reintegration.

Do not manufacture exhaustion to prove a point.

Use naturally occurring variation across realistic papers.

24. Compare What Happened Immediately Before the Target

A student can apply a repaired rule after an easy question and fail after a difficult one.

The difference may be state, not knowledge.

Did the previous question consume time?

Did the learner carry frustration forward?

Did a method switch create confusion?

Did the student rush because of a newly visible clock deficit?

Context around the target can explain why a repair appears inconsistent.

25. Compare the Cost of Success

Correctness is necessary.

It is not the only variable.

If a repaired operation takes four times longer than it should, the learner may still be vulnerable in a full paper.

Track approximate cost:

time;
working complexity;
number of restarts;
need for checking;
visible hesitation;
number of rereads;
dependence on scratch work.

A repair is stronger when it becomes both more reliable and more economical.

26. Compare the Learner’s Confidence—but Do Not Let Confidence Decide

A learner may say, “I’m fine with percentages now.”

Useful.

Then inspect the work.

High confidence with repeated fresh errors suggests calibration remains weak.

Low confidence with repeated correct independent performance suggests the learner may be stronger than they feel.

The second paper can therefore test two things at once:

Does the repaired capability work?
Does the learner know when it works?

27. The Tutor Should Define a Stop Rule for Celebration

This does not mean becoming joyless.

Celebrate improvement.

Then specify what the result earns.

For example:

The reference-quantity repair held on four fresh opportunities, including one late in the paper. We will move it to maintenance and stop spending a full lesson on it unless it returns.

That is a meaningful celebration because it changes the route.

28. The Tutor Should Also Define a Reopen Rule

A repair can be moved to maintenance and later return.

Do not treat that as betrayal.

Set a reopen rule.

If the same mechanism produces two meaningful failures across later mixed or timed work, reopen the item for diagnosis.

The exact threshold can vary.

The important idea is that a “closed” repair is not erased from system memory.

29. One Second Paper Is Not a Trend

A single follow-up paper can show successful reintegration under one later condition.

It cannot establish long-term stability across every paper, month and examination.

This matters especially when adults want certainty.

“Is it fixed?”

A careful answer may be:

It survived the first fresh full-paper reintegration. We are moving it to maintenance and watching whether it remains stable across later work.

That is not hedging.

That is evidence discipline.

30. Avoid Tiny Score Overinterpretation

Paper 1: 74.

Paper 2: 76.

Two marks may matter to a grade boundary.

They may not mean the learner’s underlying capability changed by exactly two percent.

Different papers sample different questions. Marking may vary slightly. A single lucky or unlucky item can move the total.

Use small score changes carefully.

Mechanism-level evidence is often more useful for tuition decisions.

31. Multiple Repairs Make Attribution Harder

If the tutor changes ten things between Paper 1 and Paper 2, improvement becomes harder to explain.

Sometimes ten things genuinely need changing.

But where possible, prioritise a short repair queue.

That was the principle in Volume 0022.

A short queue lets the second paper answer sharper questions:

Did the stop-loss rule reduce cascade damage?
Did reference-quantity errors fall?
Did evidence-to-inference writing survive time pressure?

32. Repair One Mechanism Without Freezing the Rest of Learning

Prioritisation does not mean the student studies nothing else.

School continues.

Homework continues.

33. Mathematics: Did the Representation Repair Survive?

Suppose the first Mathematics paper showed a recurring problem:

The learner could calculate accurately once an equation existed, but repeatedly formed the wrong equation from word problems.

The repair phase focused on defining the unknown, identifying relationships and representing the problem before calculation.

In the second paper, do not ask only whether word-problem marks improved.

Inspect the entry behaviour.

Did the learner name the unknown?
Did the equation reflect the relationship correctly?
Did diagrams or tables appear where useful?
Did representation become faster?
Did the same discipline survive after topic switching?

Now you know whether the repair travelled into the whole paper.

34. Additional Mathematics: Did the Better Method Become the Default?

A-Math students often know more than one valid method.

The first paper may show repeated selection of long, fragile routes.

After repair, the second paper should reveal whether the learner now recognises lower-cost structure without being told.

Look for:

earlier recognition of substitution, factorisation or identity structure;
fewer unnecessary expansions;
less algebraic clutter;
fewer sign-exposure points;
more time preserved for later questions.

The strongest evidence is not that the student can perform the better method when asked.

It is that the student now chooses it independently when the paper makes the choice necessary.

35. English: Did the Answering Repair Survive New Text?

Suppose the first English paper showed a repeated failure to answer the exact relationship in the question.

The learner gave relevant language but weak task alignment.

The repair phase trained:

read the command → identify the relationship → locate evidence → answer that relationship directly.

In the second full paper, the passage is fresh.

Now ask:

Did the learner distinguish cause, effect, inference, comparison and purpose?
Did evidence match the actual question?
Did time pressure shorten answers without changing their logical job?
Did the learner preserve the repair when the passage became difficult?

That is reintegration in English.

36. Science: Did the Causal Chain Survive Unfamiliar Context?

A Science learner may repair a weak explanation by learning to connect:

condition → mechanism → observable outcome.

Then the second paper changes the objects, diagram and wording.

Good.

That is where the tutor learns whether the student repaired a scientific reasoning pattern or merely memorised one corrected sentence.

Track whether the learner can:

identify the relevant condition;
retrieve the correct concept;
connect evidence to mechanism;
avoid overclaiming beyond the evidence;
write the causal chain clearly under time.

37. Primary Learners: Keep the Comparison Concrete

A Primary learner does not need the phrase “reintegration validity.”

Say:

Last paper, you kept using the wrong number as the base. We fixed that. In this paper, you chose the correct base four times by yourself. Once, near the end, you rushed and forgot. So the idea is working, but we still need it to survive when you are short of time.

The tutor keeps the deeper model.

The child gets a clear next move.

38. Secondary Learners: Teach Them to Read the Comparison

Secondary students can begin owning the mechanism map.

Ask them:

Which old error disappeared?
Where did it almost return?
What new cost did the repair create?
Which improvement held even when the question changed?
Which part still needs a reminder?
What would convince you the repair is stable?

The goal is not self-diagnosis without support.

The goal is gradually more accurate self-observation.

39. JC and Advanced Learners: Compare Trade-Offs, Not Only Errors

At advanced levels, the repair may concern resource allocation rather than a simple right-or-wrong rule.

Perhaps the learner was spending too long on high-complexity problems.

The repair introduced a stop-loss threshold.

The second paper should reveal whether the new threshold:

protected later marks;
preserved enough time for return;
did not cause premature abandonment of solvable questions;
improved total completion without destroying depth.

The repair is a policy.

The second paper tests whether that policy improves the system.

40. Class 0 · Homework Helper: Preserve the Evidence Trail

A Class 0 Homework Helper may not own the diagnosis.

But they can preserve the paper, the earlier correction, the target rule and the later evidence so nothing disappears into separate folders.

This matters because reintegration requires memory across time.

A system that forgets what it repaired cannot verify whether the repair survived.

41. Class 1 · Explainer: Check Whether Understanding Still Needs a Prompt

The Explainer asks whether the concept is now sufficiently clear to be reconstructed independently.

If the student succeeds only after hearing the first line of the explanation again, the repair has not fully reintegrated.

The Explainer may need to strengthen the conceptual representation, not simply repeat the same explanation more loudly.

42. Class 2 · Drill Builder: Check Fluency Under Competition

A Drill Builder may have made an operation fast and accurate in focused practice.

The second paper asks whether that fluency remains available while attention is shared with reading, selection, timing and other topics.

If fluency disappears under competition, the operation may need more varied retrieval rather than more identical repetition.

43. Class 3 · Diagnostic Tutor: Distinguish Failed Repair From Changed Condition

The Diagnostic Tutor protects the interpretation.

If the target error returns, ask why.

Was the original diagnosis wrong?
Was the repair too narrow?
Did a new representation expose an untrained boundary?
Did time pressure reveal insufficient fluency?
Did an upstream failure create the target error downstream?
Did the learner simply forget the rule after delay?

“It came back” is an observation.

The diagnosis still needs work.

44. Class 4 · Route Designer: Decide Whether to Maintain, Strengthen or Rebuild

The Route Designer turns the second paper into a route decision.

Evidence	Route
Repair survived cleanly	Move to maintenance; advance to next priority.
Repair survived except under one condition	Train that condition; retest locally and later globally.
Old error reduced but new cost emerged	Optimise the repair without discarding it.
Repair failed broadly	Re-open diagnosis; rebuild earlier dependency.
Evidence ambiguous	Run a small fresh discriminating check before changing the whole plan.

45. Class 5 · Performance Coach: Test Survival Under Real Load

The Performance Coach focuses on what happens when the repaired skill competes for limited time and attention.

Can the learner still:

use the stop-loss rule after a hard question;
protect checking on high-risk items;
recover after a stall;
preserve answer quality late in the paper;
select the repaired method without slowing the whole run?

The second paper is where performance coaching becomes visible.

46. Class 6 · Learning Architect: Decide Whether the System Changed

The Learning Architect asks the widest question:

Did the repair improve the learner’s overall system, or did it merely improve one local task?

Sometimes a local repair produces global benefit.

Better retrieval can free time for later sections.

Better representation can reduce downstream arithmetic errors.

Better question reading can improve several topics at once.

The architect looks for propagation.

47. Alicia: Higher Score, Same Weakness

Alicia scores 64 on her first Mathematics paper.

The main repair targets percentage reference quantities.

Her second paper is 74.

Excellent overall improvement.

But the second paper contains three percentage opportunities.

She misidentifies the base twice.

The score rose because algebra and geometry happened to go much better.

The tutor celebrates the paper.

The percentage repair stays open.

48. Beatrice: Lower Score, Stronger Repair

Beatrice scores 78 on Paper 1 and 73 on Paper 2.

The target repair was sign control.

Paper 1 contained five sign-related losses.

Paper 2 contains none.

Her lower total comes from an unusually difficult geometry cluster and one incomplete long question.

The sign repair is moved to maintenance.

The route now turns to geometry and completion.

49. Ciara: The Repair Survives Until the Clock Narrows

Ciara repaired a weak inference routine in English comprehension.

In the second paper, her first three inference answers are well aligned and evidence-based.

Then one difficult vocabulary question consumes too much time.

The final inference answer becomes vague and unsupported.

The tutor does not reteach inference from the beginning.

The repair survived until time scarcity.

The next job is to make the inference routine cheaper and protect it late in the paper.

50. Denise: The Error Disappears Because She Stops Attempting the Question

Denise previously made repeated mistakes in graph interpretation.

After repair, the second paper shows no graph interpretation errors.

But she skips both graph questions and never returns.

The old error count fell to zero.

The capability did not prove itself.

The tutor records substitution and creates a fresh low-stakes graph check before changing the diagnosis.

51. Emily: The Repair Works but Costs Too Much

Emily’s writing repair focuses on planning before drafting.

The second composition is far better organised.

But she spends fourteen minutes planning and has too little time to complete the ending properly.

The planning repair is conceptually successful.

Its cost is too high.

The tutor does not remove planning.

The tutor compresses it.

52. Faith: The Second Paper Reveals a Hidden Upstream Problem

Faith repaired weak Science explanations successfully in targeted practice.

On the second paper, her explanations collapse again.

At first this looks like failed repair.

But closer reading shows that the real failure begins earlier: she misreads the experimental condition, so the correct mechanism is never selected.

The explanation skill is not the first weak link this time.

The second paper has refined the diagnosis.

53. Three Students, Three Reintegration Decisions

In a 3-pax tutorial, the same second paper can produce three completely different route decisions.

Alicia’s repair survives and moves to maintenance.

Beatrice’s repair works only untimed, so the next work is timed integration.

Ciara’s repair fails because the original diagnosis was too narrow, so the route returns upstream.

The worksheet is shared.

The interpretation is individual.

This is one reason small-group visibility matters.

54. Learning, Studying, Teaching, Training and Improvement Read the Second Paper Differently

Layer	Second-paper question
Learning	Did the repaired capability remain available after delay and surface change?
Studying	What independent practice now deserves time?
Teaching	Does any concept still need explanation or representation repair?
Training	Under which load does the repair stop surviving?
Improvement	What does this comparison change in the route?

The same paper can therefore produce several legitimate actions without becoming confused.

55. The 90-Minute Second-Paper Review Session

0–8 minutes: Reconstruct the target. Ask the learner what was repaired and what rule or behaviour they were meant to carry into the paper.
8–18 minutes: Read the total paper. Score, completion, timing and major events.
18–30 minutes: Build the target trace. Locate every opportunity where the repaired mechanism mattered.
30–42 minutes: Compare Paper 1 and Paper 2. Recognition, selection, execution, cost, prompts and position.
42–52 minutes: Check displacement. Did the repair create a new time, checking or avoidance cost?
52–62 minutes: Fresh discriminating check. Use a small unseen task where interpretation remains uncertain.
62–72 minutes: Classify the outcome. Survived, partially survived, displaced or failed.
72–82 minutes: Route decision. Maintain, strengthen, optimise or rebuild.
82–88 minutes: Student explanation. Learner states what changed and what still fails.
88–90 minutes: Handover. One next action, one evidence target, one later recheck.

The exact timing can change.

The logic should not:

reconstruct → trace → compare → discriminate → classify → route.

56. The Reintegration Decision Sheet

Field	Tutor record
Repair claim	Exact mechanism expected to change
Paper 1 evidence	Original failure pattern
Repair route	What was taught, practised or trained
Delay	Time between repair and second full paper
Paper 2 conditions	Timing, supports, freshness, state
Target opportunities	Where the mechanism appeared
Recognition	Did learner notice the relevant structure?
Selection	Did learner choose the repaired route?
Execution	Was it accurate?
Cost	Was the route efficient enough?
Load survival	Did it survive mixing, timing, late-paper pressure?
Compensation	Was success purchased with another resource?
Outcome	Survived / partial / displaced / failed
Next route	Maintain / strengthen / optimise / rebuild

57. Failure Mode: Compare Scores and Skip the Scripts

“You improved by nine marks.”

Good.

Now inspect the target mechanism.

If the original weakness still appears repeatedly, the score gain does not close the repair.

Repair: compare the whole-paper result and the target trace separately.

58. Failure Mode: Demand an Identical Paper

A tutor waits for a magically equivalent paper before drawing any conclusion.

Learning time passes.

Repair: use reasonable comparability, document meaningful differences and keep claims proportional.

59. Failure Mode: Reuse the Same Paper and Call It Transfer

The learner remembers the item order, method and corrected answers.

Performance improves sharply.

That can show successful correction.

It does not fully establish transfer to fresh material.

Repair: use same-paper reattempt for one job and a fresh paper for reintegration.

60. Failure Mode: Help the Learner Through the Target

The tutor sees the repaired question and cannot resist helping.

The student succeeds.

The tutor has changed the evidence.

Repair: preserve the independent attempt. If intervention becomes necessary, mark exactly when support began.

61. Failure Mode: Treat Any Return of the Error as Total Failure

The learner succeeds on six fresh opportunities and fails once under severe late-paper time pressure.

“The repair didn’t work.”

Too coarse.

Repair: classify partial survival and train the condition where the failure returned.

62. Failure Mode: Treat Any Correct Answer as Proof

The learner gets one target item right by a slow, uncertain route.

“Fixed.”

Too fast.

Repair: inspect selection, cost, independence and repeated opportunities.

63. Failure Mode: Ignore the New Cost

The repair removes careless errors but adds fifteen minutes of checking.

The final section becomes incomplete.

Repair: keep the useful part of the repair and compress its operating cost.

64. Failure Mode: Close the Repair Forever

A repair survives one second paper.

The tutor deletes it from attention entirely.

Months later the weakness returns under a new format.

Repair: move successful repairs to low-cost maintenance rather than pretending learning never changes.

65. Failure Mode: Keep Testing Instead of Teaching

The second paper shows the repair failed.

The tutor immediately assigns a third full paper.

Then a fourth.

The same error keeps returning.

Repair: when the mechanism is already clear, stop rediscovering it and return to targeted teaching or training.

66. What Research Supports

Research on practice testing supports the broader idea that testing can do more than measure performance. A meta-analysis by Adesope, Trevisan and Sundararajan found that practice testing can improve later learning compared with restudy and other comparison conditions. For tutoring, that supports the use of fresh retrieval opportunities as part of learning—not merely as score collection.

Research on interleaving also matters. A systematic review by Firth, Rivers and Boyle found benefits of interleaved examples for concept learning and transfer to new items under several conditions. The practical implication is not that every worksheet should be random. It is that a repair eventually needs to survive among competing alternatives if the final task requires discrimination.

How People Learn II emphasises that learning is influenced by interacting cognitive, contextual and social conditions. That wider view matters here because a repaired capability does not operate in isolation when it returns to a full paper.

The Standards for Educational and Psychological Testing emphasise appropriate interpretation and use of test results. A tutor should therefore resist stretching one practice-paper result into a claim larger than the evidence supports.

The practical conclusion is conservative:

A second full paper is useful evidence that a repair survived one later integrated condition. Strong confidence comes from repeated, fresh, independent evidence across time and tasks.

67. What Research Does Not Justify

One improved paper does not prove permanent mastery.

One lower paper does not prove regression.

One second paper does not isolate the effect of tuition from every other influence.

A practice paper does not diagnose a medical, psychological or neurodevelopmental condition.

A tutor should not convert ordinary performance variability into dramatic labels.

The evidence should remain proportional to the decision being made.

68. Evidence and Further Reading

69. The Second Full-Paper Operating Cycle

Read Paper 1 → identify the first useful weak link → define the repair claim → teach or train the mechanism → verify locally → change the surface → mix the task → add representative timing → select a fresh comparable full paper → preserve the independent attempt → mark the whole paper → build the target trace → compare recognition, selection, execution, cost and load survival → inspect displacement and compensation → use a fresh small check where evidence is ambiguous → classify the repair as survived, partially survived, displaced or failed → maintain, strengthen, optimise or rebuild → keep successful repairs in low-cost system memory → watch later work for recurrence.

This is how the second full paper becomes more than another score.

70. Final Compression

The student finishes the second paper.

Do not begin with:

Did the score go up?

Begin with:

What did we repair, and where did that repair have to operate in this paper?

Find every target opportunity.

Did the learner recognise it?

Did the learner select the repaired route?

Did the learner execute it?

Did it cost too much?

Did it survive a changed surface?

Did it survive topic switching?

Did it survive the clock?

Did it survive late in the paper?

Did it survive after a difficult question?

Did it remain independent?

Did the student avoid the old error only by avoiding the task?

Did accuracy improve only because time cost exploded?

Did one improvement create another weakness?

Now read the score.

The score matters.

It tells you how the whole performance landed.

But the repair trace tells you whether your intervention travelled.

If the score rose and the repair survived, excellent.

Move the repair to maintenance.

If the score rose and the repair failed, celebrate the broader improvement and keep the repair open.

If the score fell and the repair survived, preserve the repair and investigate the new losses.

If the repair works only untimed, train time.

If it works only when the topic is obvious, train discrimination.

If it works early but not late, train load survival.

If it works only after reminders, rebuild independence.

If it creates a new cost, optimise.

If it fails broadly, return upstream.

Do not keep assigning full papers to rediscover the same known problem.

Teach when teaching is needed.

Drill when fluency is needed.

Diagnose when the explanation is uncertain.

Train when the skill collapses under conditions.

Retest when a claim needs evidence.

Then move on when the evidence is strong enough.

The first paper tells the tutor what broke. The second paper tells the tutor whether the repair belongs to the learner yet.

That is the discipline of the Second Full Paper.

That is Tutor Handbook Volume 0023.

Next in the series: The Tutor Handbook Vol No.0024 | The Trend — How a Tutor Decides Whether Improvement Is Stable Across Several Papers.

Small Group Tutorials

The Tutor Handbook Vol No.0023 | The Second Full Paper — How a Tutor Decides Whether a Repair Survived Reintegration