The Tutor Handbook Vol No.0050 | The Decision Record — How a Tutor Preserves Why a Learning Route Changed Before Hindsight Rewrites the Story

The Tutor Handbook · Volume 0050 · Series ID THB-0050

A tutor changes the plan.

For three weeks, the learner has been practising one route. Then a new paper arrives, the evidence changes, and the tutor decides to do something different.

Perhaps the learner no longer needs explanation. Perhaps the learner needs more retrieval. Perhaps the apparent “carelessness” is actually a sign-control problem. Perhaps the learner can execute a method but cannot select it when the topic label disappears. Perhaps the learner is ready for less help. Perhaps the tutor has realised that the original diagnosis was wrong.

The decision may be excellent.

But two months later, something dangerous happens.

Everybody remembers the outcome. Nobody remembers the information state that produced the decision.

If the learner improves, the new route begins to look obvious in hindsight. If the learner struggles, the old route begins to look as though it should never have been changed. Parents remember different parts of the conversation. The tutor remembers the most recent evidence more vividly than the older evidence. The learner remembers how the decision felt rather than why it was made. A later test result quietly becomes the reason everybody thinks they acted earlier.

A Decision Record is the smallest useful record of why a tutoring route changed, what evidence justified the change, what remained uncertain, what action was chosen, and what future evidence was expected to confirm, weaken or reverse that decision.

The point is not paperwork.

The point is intellectual honesty across time.

The previous volume, The Tutor Handbook Vol No.0049 | The Learning Claim, established a discipline: do not say more about a learner than the evidence can support.

This volume takes the next step.

Once a claim is calibrated, how does a tutor preserve the reason for acting on it before the outcome arrives and rewrites the story?

A good tutor should be able to say: This is what we believed. This is why we believed it. This is what we chose. This is what we expected to happen. This is what would make us change our mind.

Quick Read

The Decision Record is a compact record of a consequential tutoring decision made before its outcome is known.
It preserves the learner state, the current claim, the evidence, the alternatives considered, the chosen action, the expected result and the review condition.
It is not a diary of every lesson, a permanent label, or a defence brief written to prove the tutor was right.
It is especially useful when a route changes, support is increased or reduced, a diagnosis is reclassified, an intervention is stopped, a learner is released, or a problem is escalated outside ordinary tuition.
A decision can be good even when the outcome is disappointing if the choice was reasonable under the evidence available at the time.
A decision can be poor even when the outcome turns out well if it was chosen for weak reasons and succeeded mainly through luck or unrelated change.
Record conditions, not only conclusions: task difficulty, support, timing, delay, marking, topic mix and learner state matter.
Record what would have made the tutor choose differently.
Record what evidence is expected next. A decision without a return condition can become a habit.
Reversible decisions need lighter records than costly or hard-to-reverse decisions.
Class 0 to Class 6 tutors make different kinds of decisions, but every class benefits from traceable reasoning.
Learning, studying, training, education and improvement create different decision objects.
Three-student tutorials create useful natural evidence because tutor attention moves and learner independence becomes visible.
The best Decision Record is short enough to use and strong enough to reconstruct the earlier information state honestly.

1. What This Volume Owns

This volume owns tutor-side decision memory.

Its narrow job is to preserve why a consequential learning decision was made before the outcome is known.

Examples include continuing a learning route; advancing difficulty; returning to a prerequisite; changing tutor function; increasing or decreasing support; changing practice structure; moving from untimed to timed work; beginning full-paper simulation; pausing a route; restarting an old repair; releasing a learner from routine intervention; and escalating an issue beyond the present tutoring scope.

The wider logic of decisions belongs elsewhere in the eduKate ecosystem, including What Is a Decision? | How Evidence, Value and Uncertainty Become Commitment. Formal assessment cut decisions have their own owner in Assessment Decision Thresholds | Pass, Fail, Mastery & Intervention.

The Tutor Handbook applies those broader ideas to a smaller professional question: what should a tutor remember about a real learner when a real learning route changes?

2. What This Volume Does Not Own

This is not the archive of the learner’s entire history. The Tutor Handbook Vol No.0034 | The Archive owns the larger problem of preserving useful learning history without letting old labels dominate the future.

This is not claim calibration itself. Volume 0049 owns the discipline of matching the strength of a statement to the evidence condition.

This is not first weak-link diagnosis. The Tutor Handbook Vol No.0004 | Diagnostic Tutor owns that work. Nor is this the choice of the smallest intervention after a cause has been separated; The Tutor Handbook Vol No.0037 | The Branch Point owns that local selection problem.

This page owns the memory of the decision: why did the route change, given what was known then?

3. Why Outcomes Rewrite Memory

Human memory is not a recording. It is reconstruction.

Once an outcome is known, earlier uncertainty becomes hard to feel. The result pulls the story toward itself. A student who later excels can make an earlier promotion look inevitable. A student who later struggles can make the same promotion look reckless. Yet neither conclusion may be fair to the evidence that existed on the day the choice was made.

A Decision Record gives the earlier information state somewhere to live.

4. A Good Outcome Does Not Automatically Prove a Good Decision

A tutor guesses that a learner needs harder work, skips two prerequisite checks, and the learner happens to cope.

The result is good. The process may still have been weak.

Why does that distinction matter? Because the tutor will make the next decision using whatever was learned from this one. If luck is mistaken for method, a fragile decision process gets reinforced.

5. A Bad Outcome Does Not Automatically Prove a Bad Decision

Suppose a learner has three fresh pieces of evidence showing stable independent performance. The tutor reduces support. The next school paper happens to contain an unusually difficult cluster and the score falls.

The disappointing outcome does not, by itself, prove that reducing support was irrational. It tells the tutor to inspect what failed and update the learner model.

Professional judgement improves when decisions and outcomes are evaluated separately before they are recombined.

6. Start With the Decision Question, Not the Answer

A weak record begins: “We decided to add more practice.”

A stronger record begins: “Does the learner need more execution practice, or is method selection the actual bottleneck?”

The question preserves the fork. It reminds the tutor that another route was possible.

7. Name the Learner State Before the Action

A route change should begin with the state that made the old route insufficient.

“Accuracy is high on blocked algebra sets, but mixed sets still produce method-selection errors” is a state. “Needs more challenge” is mostly a conclusion.

State descriptions make later review possible because they give the tutor something specific to compare against.

8. Preserve the Learning Claim

Volume 0049 asked the tutor to calibrate claims. The Decision Record carries that claim into action.

For example: “The learner can execute the procedure accurately when the topic is named, but independent selection has not yet been demonstrated.”

That sentence is more useful than “student knows the topic” because it makes the next test obvious.

9. Preserve the Evidence That Actually Mattered

Do not copy every worksheet score into the record. Capture the evidence that changed the decision.

two fresh mixed sets;
one delayed retrieval check;
a marked school paper;
an observed failure under time;
a successful unsupported reattempt;
a repeated answer-scope error across unrelated passages;
a parent observation that was later reproduced under known tutorial conditions.

The record is not a warehouse of evidence. It is a map of which evidence carried the decision.

10. Preserve the Conditions

An 80 per cent result means something different with notes open, after a hint, under no time limit, or on a familiar worksheet.

The conditions do not need a forensic report. A few words are usually enough: fresh; unsupported; timed; delayed one week; mixed topic; marked to school standard.

Those qualifiers keep later readers from making the evidence larger than it was.

11. Preserve the Best Alternative Explanation

A serious decision record includes the strongest competing story.

If the tutor thinks slow performance is caused by weak retrieval, the competing explanation may be that the learner is spending too long checking. If the tutor thinks comprehension is weak, the alternative may be that answer scope rather than passage understanding is causing the marks to fall.

Recording one plausible alternative is often enough to prevent premature certainty.

12. Preserve Why the Chosen Action Was Preferred

The action should be connected to the claim.

“Use a mixed set because execution is already stable and the unresolved question is method selection.”

That is better than “Do mixed questions next week.”

The first version preserves the logic. The second preserves only the activity.

13. Preserve the Expected Receipt

Every meaningful intervention should produce a receipt.

If the route is right, what should become visible?

Maybe sign errors should fall while conceptual errors remain unchanged. Maybe a learner should begin tasks with fewer prompts. Maybe completion time should shorten without a drop in accuracy. Maybe a writing answer should become shorter because irrelevant evidence is being removed.

A predicted receipt turns an intervention into a testable decision.

14. Preserve the Review Condition

A route without a return condition can become permanent by inertia.

“Review after three fresh mixed sets.” “Review after the next marked school paper.” “Review once independent starts are stable for two weeks.”

The review condition can be a date, a number of samples, a performance threshold or the arrival of new evidence. It should be specific enough that the tutor knows when the decision comes back onto the table.

15. Preserve the Reversal Condition

What would make the tutor undo or change the route?

If support has been reduced, perhaps two consecutive unsupported failures on previously stable skills trigger re-entry. If timed work has begun, perhaps accuracy falling below an agreed floor sends the learner back to untimed repair. If a new study routine increases completion but destroys sleep, the route needs reconsideration.

A reversal condition is not pessimism. It is pre-committed responsiveness.

16. The Seven-Line Decision Record

Decision question: what fork are we resolving?
Current state: what is happening now?
Current claim: what does the evidence support?
Evidence and conditions: what carried the claim, and under what conditions?
Action and reason: what are we doing, and why this rather than the nearest alternative?
Expected receipt: what should change if the decision is sound?
Review and reversal: when do we look again, and what would make us change course?

For many tutoring decisions, those seven lines are enough.

17. Record Depth Should Match Decision Consequence

Not every choice deserves a full page.

Changing tonight’s drill order is cheap and reversible. Reducing a major support system before an examination is more consequential. Recommending a change of educational route is larger again.

The deeper the consequence, the stronger the need to preserve assumptions, alternatives, uncertainties and review conditions.

18. Reversible Decisions Should Stay Light

Tutoring often benefits from small experiments.

Try a shorter retrieval block. Change the order of two exercises. Remove one prompt. Mix two question families. Ask the learner to explain before calculating.

When the cost of reversal is low, the record can be brief. The discipline matters more than the length.

19. Hard-to-Reverse Decisions Need More Evidence

Some decisions allocate months of time, reshape expectations, or involve another professional or institution.

For these, the record should show more than one convenient signal. It should show the evidence window, the main alternative explanations, the reason for acting now, and the cost of waiting versus acting.

The aim is not bureaucracy. It is proportional care.

20. The Record Protects Against Tutor Overconfidence

Tutors see patterns quickly. That is often a strength.

It can also create a danger: the story becomes clear before the discriminating evidence arrives.

Writing “strongest current explanation” and naming one alternative slows the leap from pattern recognition to certainty. The record gives intuition a place, but it also gives uncertainty a place.

21. The Record Protects Against Tutor Self-Defence

When an intervention fails, the temptation is to protect the intervention.

“The student did not practise enough.” “The school paper was strange.” “The learner was tired.” Any of those may be true. But the Decision Record asks a harder question: what did we predict, and did the receipt arrive?

If not, the route deserves inspection before blame is assigned.

22. The Record Protects Against Parent Hindsight

Parents care because the stakes are real. That makes later outcomes emotionally powerful.

A simple earlier record can clarify that a decision was conditional rather than absolute: “We are trying this for three weeks because these two indicators have stabilised. We will reverse if the unsupported work deteriorates.”

That language makes the route understandable without pretending the future is known.

23. The Record Protects the Learner From Labels

Beatrice loses marks on comprehension questions.

A label says: “Beatrice is weak at comprehension.”

A decision record says: “On three evidence questions, Beatrice located the relevant line but over-expanded the answer beyond the question scope. For the next two weeks, train answer-boundary control rather than generic inference.”

The record attaches the problem to a process and a time-bounded intervention. It does not turn the problem into the child.

24. This Is Not a Behaviour Dossier

A Decision Record should not become surveillance.

It does not need every late arrival, every sigh, every unfinished page, or every parent message. Capture only what materially changed the learning decision.

Good records are selective because relevance is part of judgement.

25. This Is Not a Scoreboard

A list of marks is data, not yet a decision record.

A Decision Record explains what the tutor inferred from the marks, how comparable they were, which alternative explanations remained plausible, and what changed because of them.

The difference is reasoning.

26. This Is Not a Prediction of the Child’s Future

“She will never cope with Additional Mathematics” is not a useful decision record.

“Current algebraic manipulation is not yet stable enough for the next route; repair these prerequisites and review after three independent mixed sets” is.

The record should describe the present decision boundary, not convert it into destiny.

27. Learning Decisions

Learning decisions concern what the learner understands, retrieves, represents, reasons with and can use.

Typical Decision Records ask whether to explain again, test a prerequisite, change representation, introduce variation, delay retrieval, or move from guided to independent application.

The receipt should be a changed capability, not merely a completed activity.

28. Studying Decisions

Studying decisions concern how the learner allocates attention, time, retrieval, review and resources outside the immediate act of teaching.

A study decision might replace a beautiful weekly timetable with a rolling three-day plan because the evidence shows that adaptation, not planning knowledge, is the real bottleneck.

The receipt is whether the routine survives ordinary disruption.

29. Training Decisions

Training decisions concern reliability.

Should practice stay blocked or become mixed? Should speed be added? Should the interval increase? Should support be faded? Should the learner practise error recovery rather than only perfect runs?

The receipt is repeated performance under the condition being trained.

30. Education Decisions

Education decisions place the learner inside curricula, school stages, assessments and transitions.

The tutor may need to decide what to prioritise before a school examination, whether a learner is ready to move into a more demanding topic sequence, or whether an educational issue needs coordination with parents or school.

These records should make institutional constraints visible rather than pretending the tutorial owns the whole environment.

31. Improvement Decisions

Improvement is not one intervention. It is a loop:

Observe → Diagnose → Prioritise → Repair → Practise → Connect → Perform → Review.

A Decision Record marks the moments when that loop changes direction. It records why the learner moved from diagnosis to repair, repair to practice, practice to performance, or performance back to diagnosis.

Without those transition records, a programme can accumulate activity while losing the logic that connected it.

32. Class 0 · Homework Helper Decisions

In the eduKate Tutor Classification Model, the Class 0 Homework Helper works closest to task completion and routine.

A Class 0 Decision Record might say: “The learner now starts homework without reminders but still checks answers after every item. Reduce checking access to the end of each section and review whether completion remains accurate.”

The decision is small, but the reasoning is still visible.

33. Class 1 · Explainer Decisions

The Class 1 Explainer decides when another explanation is useful and when explanation has become a substitute for learner reconstruction.

A useful record distinguishes “the learner did not understand the explanation” from “the learner understood the explanation but cannot yet generate the idea independently.” Those states need different next moves.

34. Class 2 · Drill Builder Decisions

The Class 2 Drill Builder decides how repetition should change.

Volume is not the only variable. The tutor chooses spacing, interleaving, variation, retrieval load, accuracy floor and speed pressure.

The Decision Record should explain which reliability problem the drill is intended to solve.

35. Class 3 · Diagnostic Tutor Decisions

The Class 3 Diagnostic Tutor makes decisions about cause.

Because causal mistakes can send weeks of work in the wrong direction, the record should preserve the main competing explanations and the probe that separated them.

“We chose sign-control repair because accuracy returned when the algebraic structure was held constant but signs were simplified; method knowledge remained intact.” That is a decision others can later inspect.

36. Class 4 · Route Designer Decisions

The Class 4 Route Designer decides order.

Should the learner repair a prerequisite now or continue with the school sequence? Should the next month prioritise breadth or depth? Should the tutor protect a stable skill with maintenance while concentrating resources elsewhere?

Route decisions often have opportunity costs, so the record should say not only what was chosen, but what was deliberately postponed.

37. Class 5 · Performance Coach Decisions

The Class 5 Performance Coach decides how to convert available knowledge into reliable output under pressure.

Typical choices concern timing, paper order, skip-and-return rules, checking budgets, recovery after a difficult item, and the point at which simulation becomes useful.

The record should make clear whether the bottleneck is knowledge, access, selection, speed, endurance or recovery.

38. Class 6 · Learning Architect Decisions

The Class 6 Learning Architect integrates the widest system.

A decision may combine school evidence, tutorial performance, study routines, examination timing, learner self-report and parent observations. The larger system makes traceability more important, not less.

When many signals are combined, the record should identify which ones actually changed the route.

39. Tutor Classification Is a Decision Aid, Not a Student Label

The same learner may need a Class 3 Diagnostic Tutor function for one bottleneck, a Class 2 Drill Builder function for another, and a Class 5 Performance Coach function close to an examination.

The classification describes the work the tutor is doing. It should not become a permanent category attached to the child.

40. The Three Modes of Tuition Need Decision Records Too

eduKate’s Three Modes of Tuition distinguish Repair, Alignment and Frontier work.

A learner can move between these modes. A Decision Record should say why.

Repair begins because a weak link is constraining downstream performance. Alignment becomes appropriate when the learner is stable enough to travel with the school route. Frontier work begins when the learner can use spare capacity to extend beyond immediate requirements without destabilising the core.

Mode transitions should be evidence decisions, not mood decisions.

41. Three-Student Tutorials Create Natural Decision Evidence

In a three-student tutorial, the tutor’s attention moves.

That movement creates natural information. Does the learner continue when the tutor turns away? Can they mark uncertainty and wait? Do they restart after an error? Do they copy another learner’s route or preserve their own reasoning?

The small-group environment can reveal control that constant one-to-one prompting may hide. Those observations can legitimately change a support decision when they are repeated and relevant.

42. Peer Comparison Is Context, Not Identity

Three learners sitting together will differ in speed, prior exposure and confidence.

Peer comparison can help a tutor notice that a task is unusually slow for the group or that one learner has discovered a more efficient representation. It should not become the basis for a permanent claim that one child is “slow” or “weak”.

The Decision Record should return to the learner’s own evidence.

43. Alicia · Mathematics: From Execution to Selection

Alicia is accurate on blocked algebra questions. The tutor could simply raise the difficulty.

But on two mixed sets Alicia pauses until the topic family becomes obvious.

The record reads: “Execution is currently stable when the method family is signalled. Independent selection remains untested. Before increasing algebraic difficulty, use short mixed sets without topic labels. Expect faster route selection without a loss of accuracy; review after three fresh sets.”

That is a small decision with a clear receipt.

44. Beatrice · English: From Generic Comprehension to Answer Boundary

Beatrice often loses marks in comprehension. A generic response would be more comprehension practice.

Across three passages, however, she locates the relevant evidence correctly and then adds material the question did not ask for.

The decision record says: “Pause generic inference drilling. Train evidence-to-answer boundary because source selection is mostly correct while scope control is unstable. Expect shorter answers with equal or better marks. Review on two unseen passages.”

The record prevents “weak comprehension” from swallowing a more specific problem.

45. Ciara · Science: From Memorised Wording to Conditional Transfer

Ciara remembers model explanations beautifully. On a changed setup, she repeats the stored answer even when one condition has changed.

The record says: “Recall is strong; conditional transfer is not. Replace repeated model-answer rehearsal with one-variable-at-a-time scenario changes. Ask Ciara to rebuild the causal chain before writing the explanation. Review whether wording now changes appropriately with the mechanism.”

The action follows the evidence rather than the appearance of fluency.

46. Denise · Additional Mathematics: From Knowledge Repair to Performance Coaching

Denise solves calculus accurately when untimed but leaves late-paper questions incomplete.

The Decision Record says: “Current evidence supports usable calculus knowledge under untimed conditions. The main unresolved bottleneck is whole-paper time allocation and recovery after long items. Do not reteach the full calculus sequence. Train paper navigation and timed checkpoints. Review completion distribution as well as total score.”

The tutor function changes from Explainer or Drill Builder toward Performance Coach because the learner state changed.

47. Emily · Studying: From Beautiful Plans to Adaptive Control

Emily writes elegant weekly timetables every Sunday. By Wednesday they are usually obsolete.

The Decision Record says: “Planning knowledge is present; replanning after disruption is weak. Replace the seven-day fixed schedule with a rolling three-day priority plan. Expected receipt: unfinished tasks are consciously reallocated rather than abandoned. Review after two school weeks.”

The decision targets the control failure, not the appearance of organisation.

48. Faith · Transition: From Tutor Confirmation to Independent Starts

Faith’s current content is stable, but she waits for the tutor to confirm the first step of unfamiliar tasks.

The Decision Record says: “Knowledge appears sufficient for the present level; initiation remains support-dependent. Reduce first-step confirmation. Ask Faith to state her planned first move and begin before feedback. Review whether independent starts remain accurate across mixed tasks.”

The route is not more content. It is more ownership.

49. Record What Was Not Chosen

A good decision becomes clearer when the nearest rejected alternative is named.

“Did not add more blocked practice because execution was already stable.” “Did not increase tutoring frequency because the failure appeared only under timed full-paper conditions.” “Did not escalate to a broader concern because the pattern disappeared when task language was clarified.”

This is not about proving other choices were foolish. It shows what trade-off the tutor consciously made.

50. Every Decision Spends a Learning Budget

The Tutor Handbook Vol No.0047 | The Learning Budget established that time, attention, practice and recovery are finite.

A Decision Record should therefore expose opportunity cost.

If forty minutes are moved into algebra repair, what is not happening during those forty minutes? If two weeks are spent on full-paper simulations, which content expansion is being postponed?

Good decisions are not only about benefit. They are about benefit relative to what the learner gives up.

51. The Record Must Respect the Scope Boundary

The Tutor Handbook Vol No.0048 | The Scope Boundary asks what belongs inside tuition and what must be coordinated or referred elsewhere.

A Decision Record must preserve that distinction.

“Academic performance declines sharply after sustained reading and the pattern persists across several sessions” may justify adapting the learning environment and discussing the pattern with parents. It does not authorise the tutor to invent a medical or psychological diagnosis.

Record what was observed, what educational action was taken, and where ownership changes.

52. Preserve Uncertainty in the Record

Useful phrases include:

“strongest current explanation”;
“not yet tested under time”;
“support conditions partly unknown”;
“school-paper difficulty may also explain part of the change”;
“working hypothesis”;
“review before broadening the claim”.

Uncertainty is not weakness when it is attached to a plan for reducing it.

53. A Decision Record Is a Snapshot, Not a Verdict

Date the record.

The statement “current best explanation is retrieval failure” belongs to a moment in the learner model. New evidence can replace it.

Versioning matters because a good learning system expects its own explanations to improve.

54. The Fresh Look Rule

The Tutor Handbook Vol No.0035 | The Fresh Look warns against assuming that a familiar symptom has the same cause as before.

The Decision Record helps because the old explanation is preserved precisely. The tutor can reuse history without being trapped by it.

“Last time this pattern came from sign control” is useful prior information. It is not proof that today’s pattern has the same cause.

55. The Differential Rule

Volume 0036 | The Differential separates old cause, new cause and mixed cause.

When that separation changes the route, the Decision Record should capture the discriminating evidence. Otherwise, a later reader sees only that “the plan changed” and loses the reason the tutor had confidence to change it.

56. The Branch Point Rule

Once the cause has been separated, the tutor reaches a branch point.

The Decision Record should name why the smallest useful intervention was chosen. If a one-step prompt repair is enough, do not pretend a full rebuild was necessary. If the prerequisite is broken, do not hide that fact behind more top-layer practice.

57. The Dose Rule

Volume 0038 | The Dose asks how much help to give before support begins replacing learning.

A dose change deserves a receipt. If prompts are increased, which specific failure should reduce? If prompts are reduced, what independent behaviour should appear?

Without that record, “more help” can quietly become the permanent default.

58. The Fade Rule

Volume 0039 | The Fade removes support deliberately.

The Decision Record should say which support is being removed, why now, and what collapse signal would trigger restoration. That makes fading a controlled experiment rather than an optimistic withdrawal.

59. The Independence Rule

Volume 0040 | The Independence Test asks whether unsupported success is stable self-regulation or only a temporary good run.

The Decision Record should preserve the evidence window used to cross that boundary. “Three independent starts across two weeks, including one changed task” is more useful than “seems independent now”.

60. The Release Rule

Volume 0041 | The Release asks when routine intervention can end.

Release is exactly the kind of decision hindsight can distort. If the learner later succeeds, the release looks obvious. If the learner later struggles, it can look irresponsible.

Preserve the release evidence and the check-in plan while both are still fresh.

61. The Re-Entry Rule

Re-entry should not occur because one disappointing mark creates panic.

The earlier release record tells the tutor which signals matter. The re-entry record then shows which of those signals actually appeared, whether the old cause still fits, and why structured support is justified again.

This preserves continuity without turning continuity into repetition.

62. The Stability Window Rule

Volume 0046 | The Stability Window warns against changing a new route so quickly that nobody can learn whether it works.

The Decision Record should therefore include the minimum observation window. Unless a safety or severe deterioration signal appears, hold the route long enough for the predicted receipt to become visible.

63. Decision Records and the Learning Budget

A learning budget asks where scarce capacity should go. A Decision Record explains why one destination won.

That pairing is powerful because later review can ask two questions separately: Did the intervention work? And was it worth what it displaced?

An effective intervention can still be a poor allocation if a higher-value bottleneck was neglected.

64. Decision Records and the Learning Claim

The two volumes fit together as a chain:

Evidence → Calibrated Claim → Bounded Decision → Action → Receipt → Review → Updated Claim.

If any link disappears, the system becomes harder to learn from. Evidence without a claim is noise. A claim without a decision is inert. A decision without a receipt is hard to test. A receipt without review is wasted information.

65. A Parent-Facing Decision Record Can Be Four Lines

What we are seeing: the current pattern.
What we think it means: the best current explanation.
What we are changing: the next route and why.
What we will look for next: the receipt and review condition.

That is usually enough to keep parents informed without drowning them in lesson notes.

66. A Student-Facing Decision Record Can Be Even Simpler

What is working?
What is still breaking?
What are we changing?
What should you notice if the change works?

Older learners can be invited to challenge the tutor’s interpretation. Their self-report is not automatically decisive, but it can reveal information the task record cannot.

67. Let the Learner Predict the Review Condition

Ask: “What should be easier in two weeks if this plan is working?”

This is a small metacognitive move. It shifts the learner from receiving a plan to understanding the causal purpose of the plan.

Over time, the learner begins to make better decisions about their own study because interventions stop looking like arbitrary instructions.

68. Feedback Should Change the Next Decision

The Education Endowment Foundation’s work on feedback and independent learning emphasises feedback that helps learners understand current performance and what to do next.

For the tutor, feedback becomes more powerful when it changes the next decision rather than merely commenting on the last task.

A Decision Record captures that conversion: evidence from the previous performance becomes an explicit change in the learning route.

69. Monitoring Progress Should Change Teaching

The Australian Education Research Organisation’s Monitor Progress guidance focuses on using assessment information over time to understand learner progress and adapt teaching.

The Tutor Handbook implication is simple: collecting evidence is not the endpoint. The evidence should enter a decision loop.

The Decision Record shows when and why monitoring produced a change in action.

70. Evidence-Informed Decisions Still Need Context

AERO’s Evidence Decision-Making Tool for Practitioners is built around bringing evidence into professional decisions rather than treating evidence as an automatic answer.

Tutoring requires the same discipline. Research can inform what kinds of practice are promising, but the tutor still has to decide whether a particular learner, at a particular moment, under particular constraints, should receive that intervention now.

The Decision Record makes the local reasoning visible.

71. Metacognition Turns External Decision Rules Into Internal Ones

The Education Endowment Foundation’s Metacognition and Self-Regulated Learning guidance emphasises planning, monitoring and evaluating within subject learning.

A mature learner eventually carries a lightweight Decision Record internally.

“I keep missing these because I cannot retrieve the formula, not because I cannot use it.” “My plan failed because I underestimated the workload, so I need to reallocate.” “I can do this with notes, but I have not tested unaided recall.”

That is self-regulation becoming more precise.

72. The Decision Review

When the review condition arrives, ask:

What outcome occurred?
Did the predicted receipt appear?
Did the targeted mechanism change?
Did nearby untargeted areas change?
What stayed unchanged?
Was the original claim strengthened, weakened or split into two claims?
Has an alternative explanation become more plausible?
Should the route continue, advance, be reclassified or stop?

This review reconnects the outcome to the decision without letting the outcome erase the original reasoning.

73. Judge the Process Before You Know the Next Answer

One of the best habits in decision-making is to define what would count as a sound process before the outcome arrives.

Did we use relevant evidence? Did we preserve support conditions? Did we consider a plausible alternative? Did we choose a proportionate intervention? Did we set a review point?

Those questions can be answered independently of whether the next mark happens to rise.

74. Watch for “It Was Obvious”

When a learner improves, people often say the right route was obvious.

The Decision Record can show that two explanations were genuinely plausible and that one discriminating probe tipped the balance. That is useful knowledge because it teaches the tutor how uncertainty was actually reduced.

75. Watch for “I Knew It”

After a poor outcome, “I knew it” can appear from tutors, parents or learners.

Sometimes someone really did raise the concern. Then the record should show it and future decisions should learn from that signal.

Sometimes the concern was vague, low-confidence or contradicted by stronger evidence at the time. The record protects everybody from rewriting uncertainty as certainty.

76. Watch for Decision Drift

A route can change slowly without anyone consciously deciding.

Five minutes of extra prompting becomes twenty. One extra worksheet becomes the new weekly default. A temporary focus on one topic consumes a month.

The Decision Record catches drift by asking a simple question: when did this become the plan, and what evidence justified it?

77. Watch for Threshold Drift

A tutor may initially require three independent successes before reducing support, then become impatient and release after one. Or the opposite: the learner meets the agreed criterion, but the tutor keeps adding new requirements because letting go feels risky.

Pre-recorded review criteria make those threshold changes visible. The tutor can still revise the threshold, but should say why new evidence justifies the revision.

78. Do Not Record Only Failures

If records appear only when something goes wrong, the learner history becomes distorted toward weakness.

Record major positive transitions too: support successfully faded; mixed selection stabilised; a learner began self-correcting; a recurring weak link retired; a full paper became stable enough for release.

Decision memory should preserve growth as carefully as difficulty.

79. Do Not Record Only Tutor Decisions

Learner choices are evidence of self-regulation.

A student who says, “I am going to do the mixed questions first because that is where I still need to choose the method,” is making a learning allocation decision.

When learners begin making good route decisions without adult control, that is part of the independence story.

80. Full Decision Record Template

Decision ID
Date
Learner or group
Decision question
Current state
Current learning claim
Evidence
Evidence conditions
Alternative explanation
Options considered
Chosen action
Reason for choice
Opportunity cost
Support level
Expected receipt
Review date or condition
Reversal or escalation condition
Owner of the next action
Important unknowns

Most daily decisions need only a subset. The full template exists for decisions with enough consequence to justify the extra clarity.

81. The Thirty-Second Version

State. Claim. Action. Why. Next evidence.

Example: “Fraction procedure accuracy is stable on blocked sets, but mixed selection is untested. Move to a short mixed set because execution no longer appears to be the bottleneck. Next evidence: whether the learner chooses the correct method without a topic cue.”

That takes less than a minute and preserves the important reasoning.

82. The One-Sentence Version

Because independent delayed retrieval is stable but mixed selection remains untested, move from blocked practice to a short mixed set and review whether method choice survives without topic labels.

A good one-sentence record contains a condition, a decision and a receipt.

83. What Not to Write

Lazy.
Careless.
Weak student.
Not motivated.
Bad at English.
Needs more practice.
Not ready.
Cannot cope.

These phrases are usually too compressed to carry a good decision. Replace them with an observable process, a bounded claim, and the next test.

84. The Professional Standard

A professional tutor should be able to reconstruct the important route changes in a learner’s programme without pretending that every decision was perfect.

The goal is not to create an immaculate history in which the tutor is always right.

The goal is to create a learnable history in which good reasons are preserved, weak reasons can be identified, uncertainty is visible, and future decisions become better because earlier decisions left receipts.

85. Connected Reading Across the eduKate Ecosystem

86. Evidence and Practice Foundation

These sources support broad principles used here: monitor progress over time, use evidence to adapt teaching, build planning-monitoring-evaluation into learning, and make feedback actionable. They do not prescribe the Tutor Handbook’s exact record format. The Decision Record is eduKate’s practical tutor-side implementation of those principles inside a learning route.

Final Compression

Do not wait for the outcome to decide what you thought before the outcome.

Name the decision question.

Describe the learner state.

State the calibrated claim.

Preserve the evidence and its conditions.

Name the strongest alternative explanation.

Choose the smallest useful action.

Say why this action beats the nearest alternative.

Notice what the choice costs.

Predict the receipt.

Set the review condition.

Set the reversal condition.

Act.

Read the receipt.

Do not defend the old plan merely because it was yours.

Do not punish the old plan merely because a later outcome disappointed you.

Update the claim.

Update the route.

The Decision Record is not there to prove that the tutor was right. It is there to preserve what was knowable at the time, make the next decision traceable, and help the tutor, learner and parent become better at changing course for the right reasons.

That is the Decision Record.

That is Tutor Handbook Volume 0050.

Small Group Tutorials