Tutor Handbook Vol 0004 | Diagnostic Tutor

The Tutor Handbook · Volume 0004 · Series ID THB-0004

A student gets the same kind of question wrong for the fourth time.

The obvious response is more practice.

Ten more questions. Another worksheet. Another explanation. Another round of corrections. Another hour.

Sometimes that is exactly right.

Sometimes it is the educational equivalent of pressing harder on the wrong button.

A Diagnostic Tutor does not begin by asking, “What can I give this student next?” The Diagnostic Tutor begins by asking, “What is the earliest useful point at which this learner’s performance becomes unreliable?”

That change in question is the heart of Class 3 in the eduKateSG Tutor Classification Model.

Class 0 helps the learner complete work. Class 1 explains what is unclear. Class 2 builds fluency through designed practice. Class 3 arrives when those jobs are not enough because the visible mistake may not be the real problem.

This volume is not a replacement for eduKate Sengkang’s canonical guide How Learning Diagnosis Works. That page owns the general mechanism of diagnosis. This handbook has a narrower operational job: what must a tutor actually do when taking responsibility for Class 3 diagnostic work with a real learner?

Quick Read

A Diagnostic Tutor separates symptoms from causes before prescribing more work.
One score, one mistake or one bad day is not a complete diagnosis.
The tutor begins with observable evidence, not labels such as “careless”, “lazy”, “weak” or “bad memory”.
Several plausible explanations should remain alive until a small discriminating test separates them.
The earliest useful weak link is not always the oldest gap. It is the earliest repairable dependency that unlocks meaningful downstream performance.
Diagnosis is incomplete until the proposed repair is tested on fresh work, after reduced support and, where useful, after a delay.
A good Diagnostic Tutor knows when the problem belongs to another tutor class, a teacher, a parent conversation or an appropriate professional outside tuition.
The long-term target is greater independent capability, not permanent dependence on diagnosis.

1. The Same Mark Can Hide Different Problems

Imagine four students all scoring 55 marks.

Alicia understands most of the Mathematics but loses marks because she chooses the wrong method when several methods are possible.

Beatrice knows the methods but reads too quickly and repeatedly misses a condition in the question.

Ciara can solve familiar examples but becomes dependent on chapter labels and worked examples. In a mixed paper, she does not recognise what she knows.

Denise has a genuine foundational gap. The present topic keeps collapsing because an earlier idea was never made stable.

The mark is identical.

The next lesson should not be.

This is why diagnosis matters. A score tells us that a performance happened under particular conditions. It does not automatically identify the mechanism that produced the performance.

The Diagnostic Tutor reads beneath the total.

2. What Is a Diagnostic Tutor?

In the eduKate tutor classification, the Diagnostic Tutor is Class 3: the Root-Cause Repair Tutor.

The role begins when the tutor must decide why performance is breaking before deciding what to teach or practise next.

A Diagnostic Tutor is therefore not defined by giving harder tests, using more data or sounding analytical. The defining capability is discrimination.

Can the tutor distinguish between different causes that produce a similar visible error?

If the answer is yes, tuition can become selective.

If the answer is no, the tutor may simply add volume to uncertainty.

3. Diagnosis Is a Decision Process, Not a Label

Weak diagnostic language turns a temporary observation into a permanent description of the child.

“She is careless.”
“He has poor memory.”
“She is weak in Science.”
“He cannot do algebra.”
“She panics in exams.”
“He is lazy.”

These statements may contain a useful suspicion. They are not yet diagnoses.

A Diagnostic Tutor rewrites them into observable form.

Label	Observable rewrite
Careless	Skipped units on four of six multi-step questions despite correct calculations
Poor memory	Could explain the concept with notes open but could not retrieve the first step after two days
Weak in algebra	Could simplify expressions but could not translate word relationships into equations
Bad at Science	Could state facts but explanations did not connect evidence to mechanism and outcome
Panics	Untimed performance was stable; errors increased sharply after the learner began monitoring the clock

Specific observations can be tested.

Labels tend to defend themselves.

4. The Diagnostic Tutor’s First Discipline: Do Not Prescribe Too Early

Students usually arrive with a proposed solution already attached to the problem.

“She needs more practice.”

“He needs to memorise the formulas.”

“She needs more difficult papers.”

“He needs someone to explain everything again.”

These may eventually be correct. But the Diagnostic Tutor temporarily separates the proposed remedy from the evidence.

The working rule is simple:

Describe first. Explain second. Intervene third.

That sequence protects the learner from receiving a treatment for a problem that has not yet been located.

5. The Core Diagnostic Loop

The Class 3 operating loop can be compressed into one line:

OBSERVE → BOUND → DECOMPOSE → HOLD MULTIPLE EXPLANATIONS → DISCRIMINATE → LOCATE → REPAIR → RETEST → REDUCE SUPPORT → RETURN LATER → UPDATE.

Each verb matters.

Stage	Diagnostic question
Observe	What actually happened?
Bound	Where does the problem appear and where does it not?
Decompose	What smaller operations make up this performance?
Hold explanations	What different causes could produce this same symptom?
Discriminate	What small change would separate those causes?
Locate	Where is the earliest useful weak link?
Repair	What is the smallest appropriate intervention?
Retest	Did the suspected mechanism improve?
Reduce support	Does the learner still succeed with less help?
Return later	Did the capability remain available?
Update	Does the new evidence support or revise the diagnosis?

6. Start With the Work Itself

A marked paper is often more useful than a long story about the marked paper.

The story matters. So does the student’s experience. But the artifact preserves details that memory tends to compress.

Which questions were attempted?
Where did working change direction?
Which wrong answers were internally consistent?
Which errors repeat?
Which question types are strong?
Where did the learner erase or restart?
Did the student choose an appropriate representation?
Were mistakes conceptual, procedural, interpretive or checking-related?
Was the final answer wrong even though the method was mostly sound?

This is one reason eduKate Sengkang often asks parents to bring marked work to a consultation. A total score is useful. The written path can be more diagnostic.

7. Bound the Problem Before Expanding It

“Weak in Mathematics” is too large to diagnose.

“Struggles with algebra” may still be too large.

“Forms equations incorrectly when the unknown is described indirectly in word problems, but solves equations accurately once they are formed” is much more useful.

Bounding means finding the edge of the problem.

Does it happen only in one subject?
Only under time?
Only when the question is unfamiliar?
Only when several steps must be held together?
Only when the learner has to select the method?
Only after a delay?
Only when support disappears?
Only in written work but not oral explanation?

A bounded problem is smaller, safer and easier to test.

8. Decompose Performance Into Its Actual Jobs

A single exam question may contain many jobs.

In Mathematics:

Read → identify quantities → recognise relationships → represent → select method → execute → track units → check reasonableness → present.

In English comprehension:

Interpret the question → locate evidence → distinguish literal from inferred meaning → connect evidence to demand → phrase precisely → check completeness.

In Science:

Read conditions → identify the scientific object or relationship → separate observation from inference → retrieve the relevant concept → explain the mechanism → connect mechanism to condition → state outcome → check against evidence.

If the tutor sees only “right” and “wrong”, all of these internal jobs disappear.

Diagnosis begins when the chain becomes visible.

9. Hold More Than One Explanation Alive

Premature certainty is one of the main dangers in diagnostic tutoring.

Suppose Emily is slow on algebra questions.

Possible explanations include:

weak arithmetic fluency;
slow retrieval of algebraic rules;
uncertain method selection;
overchecking;
poor layout that forces repeated rereading;
difficulty holding intermediate results;
genuine conceptual uncertainty;
careful but accurate working that is simply slower than the tutor expected.

The observation “slow” does not choose among them.

A Diagnostic Tutor resists the temptation to turn the first plausible story into the final one.

10. The Discriminating Test: Change One Thing That Matters

Diagnosis improves when the tutor changes one condition that should matter differently under competing explanations.

If we suspect retrieval rather than understanding, close the notes and ask for the first step.

If we suspect representation, give the same relationship once in words and once as a diagram.

If we suspect prompt dependence, remove one familiar hint.

If we suspect time pressure, compare a short timed set with an otherwise similar untimed set.

If we suspect vocabulary, keep the underlying concept constant while simplifying the language.

If we suspect a conceptual gap, ask for an explanation in a simpler example before returning to the original problem.

The best diagnostic test is not the biggest test. It is the smallest change that can change the decision.

11. Diagnostic Tutoring and Formative Assessment

Diagnostic tutoring sits close to formative assessment because both use evidence to decide what should happen next.

The Education Endowment Foundation’s current Embedding Formative Assessment materials describe classroom questioning and tasks as ways to gather evidence of learning, alongside clear learning intentions, feedback and greater learner ownership.

The connection is important, but the roles are not identical. Formative assessment is a broad instructional process. Class 3 diagnostic tutoring is a specific tutor capability: when evidence is ambiguous, identify which underlying operation should be repaired before selecting the intervention.

The Diagnostic Tutor therefore asks not only, “What does the student know?” but, “What does this particular performance allow us to infer, and what remains uncertain?”

12. Feedback Is Data for the Next Decision

Feedback is not merely something the tutor gives.

It is also something the tutor reads.

If the tutor says, “Look at the unit,” and the learner immediately repairs the answer, that tells us something different from a learner who remains unable to proceed.

If one worked example unlocks five later questions, the barrier may have been route visibility. If five worked examples are needed and the learner still cannot choose the method independently, the problem is different.

The EEF’s Feedback evidence summary emphasises feedback that gives specific information about the task, subject and self-regulation. For the Diagnostic Tutor, the learner’s response to that feedback becomes additional evidence.

Feedback changes the learner.

The way the learner changes helps refine the diagnosis.

13. The Earliest Useful Weak Link Is Not Necessarily the Oldest Gap

Students have many imperfections.

A Diagnostic Tutor does not excavate every one of them.

The phrase first useful weak link matters because “first” is functional, not archaeological.

A Secondary 3 learner may have forgotten a Primary 5 fact. If that fact is irrelevant to the present difficulty, rebuilding it first may waste time.

A different learner may have a tiny gap in fractions that repeatedly destroys algebra, ratio and percentage work. That earlier dependency may be worth repairing immediately because one repair unlocks several downstream jobs.

The Diagnostic Tutor therefore looks for leverage:

Which repair is early enough to matter, specific enough to teach and powerful enough to improve more than the single question in front of us?

For the deeper canonical explanation, see Finding the First Weak Link in Learning.

14. Mathematics: Diagnose the Decision Before the Calculation

Alicia is accurate when practising equations chapter by chapter. In mixed questions, she repeatedly chooses the wrong method.

A weak intervention gives her another fifty equations.

A Diagnostic Tutor asks what those fifty equations would actually train. If the worksheet announces the method through its chapter heading, Alicia may become faster at equation solving while remaining weak at method selection.

So the tutor runs a discriminating sequence:

Give one familiar equation. Alicia solves it correctly.
Give a word problem and ask her to represent the relationships without solving. The representation is wrong.
Give the same relationship as a diagram. She represents it correctly.
Return to the wording and ask which phrase carries the relationship. She hesitates.

Now “weak at algebra” has become a more useful hypothesis: the first break may sit in translating relational language into a mathematical representation.

The repair should target that operation.

Only then should the tutor return to mixed algebra and see whether selection improves.

15. English: Diagnose Before Correcting the Sentence

Beatrice writes a weak comprehension answer.

The sentence is vague. It would be easy to rewrite it for her.

The Diagnostic Tutor does not begin with the sentence.

The tutor asks:

Did Beatrice understand the question demand?
Did she locate relevant evidence?
Did she interpret that evidence correctly?
Did she understand the relationship between evidence and answer?
Was the idea correct but the phrasing imprecise?

If the evidence choice is wrong, polishing the sentence teaches at the wrong layer.

If the evidence is correct and the idea is clear orally, then expression may genuinely be the weak link.

This distinction matters because language work is particularly vulnerable to invisible adult substitution. A tutor can easily improve the final sentence while learning very little about the learner’s independent reading process.

16. Science: Separate Knowledge From Explanation

Faith knows the topic words.

She can define evaporation, heat transfer, force, adaptation and photosynthesis when asked directly.

Yet application questions remain weak.

The tempting conclusion is that she needs more facts.

The Diagnostic Tutor tests the chain instead.

Ask Faith what changed in the scenario.
Ask which concept could explain that change.
Ask what mechanism connects the concept to the observed outcome.
Ask which detail from the question must appear in the answer.

Suppose Faith retrieves the right concept but cannot connect the mechanism to the question’s condition.

That is not a fact-recall problem.

More flashcards may increase vocabulary without repairing the explanatory operation.

The tutor should teach the causal link, then test it on a changed example.

17. When the Learner Can Do It With Help but Not Alone

This is one of the most important diagnostic states.

Ciara looks strong during tuition. She answers when the tutor asks guiding questions. She follows worked examples. She completes practice accurately.

Then the school test arrives and the capability disappears.

It is easy to say, “She knows it but cannot perform in exams.”

That may be true. It may also be incomplete.

The tutor must ask what support was present during tuition:

Was the topic label visible?
Was the tutor asking the first question?
Was the relevant formula already on the page?
Were questions grouped by method?
Was checking immediate?
Could the learner ask for confirmation at every step?

Sometimes the missing capability is not “knowing the method”. It is selecting, initiating or checking the method without environmental cues.

That is why support conditions are part of diagnosis.

18. Remove Support Carefully to Reveal the Real State

A Diagnostic Tutor sometimes learns more by removing one support than by adding another.

Remove the chapter heading.

Cover the worked example.

Delay the hint.

Ask the learner to choose the first step.

Move from blocked practice to a mixed set.

Return to the concept two days later.

The purpose is not to catch the learner failing.

It is to discover which parts of the performance have become portable.

eduKate Sengkang develops this wider question in How We Know Learning Has Really Held: learning becomes more convincing when it survives reduced support, changed conditions and delayed return.

19. One Good Question Can Be More Diagnostic Than Twenty Routine Questions

A diagnostic question should be chosen for information, not volume.

Twenty routine questions can show that a student can repeat a familiar procedure.

One carefully changed question may reveal whether the learner understands why the procedure applies.

For example:

Change the representation but preserve the underlying relationship.
Change irrelevant surface details but preserve the mechanism.
Remove the cue that normally announces the method.
Ask the learner to explain why a tempting wrong method fails.
Ask for a counterexample.
Ask what would have to change for the answer to become different.

These questions create contrast.

Contrast is diagnostic because it reveals what the learner is actually attending to.

20. The Tutor Must Diagnose the Task Too

Not every poor result belongs to the learner.

A question may be ambiguous. A worksheet may contain an error. A marking key may accept one narrow formulation when several are defensible. A practice paper may be much harder than the learner’s normal school work. A digital platform may count a formatting issue as a conceptual mistake.

The Diagnostic Tutor therefore separates:

Learner → task → support → performance → marking → interpretation.

If any of those layers changes, the meaning of the result may change.

This prevents an educational system from turning every inconvenient score into a defect inside the student.

21. Do Not Confuse Assessment With Diagnosis

An assessment produces evidence.

Diagnosis interprets evidence for a decision.

A test can be useful without being diagnostic by itself.

The U.S. Institute of Education Sciences and What Works Clearinghouse publish evidence-based practice guides for learners who struggle in Mathematics, including the 2021 guide Assisting Students Struggling with Mathematics: Intervention in the Elementary Grades. Its recommendations include systematic instruction, clear mathematical language and carefully chosen representations.

The practical lesson for a Diagnostic Tutor is broader than Mathematics: evidence should change the instructional choice. Collecting more scores without changing the decision is not automatically better diagnosis.

22. When More Practice Is the Correct Diagnosis

Class 3 should not become suspicious of repetition.

Sometimes the diagnosis is genuinely simple:

the learner understands the concept;
selects the correct method;
can explain why it works;
makes no systematic conceptual error;
but execution remains slow, effortful or inconsistent.

That is a Class 2 handoff.

The Diagnostic Tutor has succeeded by showing that the learner does not need another explanation or a deeper reconstruction. The learner needs fluency.

Diagnosis should simplify intervention when possible.

It should not make every case complicated.

23. When Explanation Is the Correct Diagnosis

Sometimes the learner simply does not understand the concept.

The tutor tests a simpler version. The misconception remains. The learner cannot explain the relationship, cannot recognise a correct example and cannot use the idea even without time pressure.

That is a Class 1 handoff.

The Explainer now owns the next move.

For the mechanism of making invisible structure visible without doing the thinking for the learner, see How Explanation Works in Teaching.

Good diagnosis does not compete with explanation.

It tells us when explanation is the right tool.

24. When the Problem Is Route Design

Denise has five real gaps.

All five deserve repair.

The new question is order.

Which comes first?

Which can wait?

Which is foundational?

Which is exam-critical?

Which repair will reconnect her to current schoolwork fastest?

Now the job has moved toward Class 4: Route Designer.

Diagnosis identifies the broken parts. Route design decides how the repair programme should unfold across time.

25. When the Problem Appears Only Under Examination Conditions

A learner may be stable in ordinary practice and still underperform in an examination.

The Diagnostic Tutor should first verify the contrast.

Does the same knowledge work untimed?
Does accuracy change as the paper progresses?
Does method selection deteriorate under mixed conditions?
Are errors clustered after difficult questions?
Does the learner spend too long seeking certainty before moving on?
Does checking disappear as time falls?

If the underlying subject capability is genuinely stable and the remaining difficulty is output under pressure, the role moves toward Class 5: Performance Coach.

Class 3 should diagnose the boundary accurately rather than absorbing every examination problem into general weakness.

26. The Diagnostic Tutor Must Know What Not to Infer

Tutors observe learning behaviour.

That does not make ordinary tuition evidence sufficient for clinical diagnosis.

A learner who loses concentration is not thereby diagnosed with ADHD. A learner who struggles to read is not thereby diagnosed with dyslexia. A learner who worries before tests is not thereby diagnosed with an anxiety disorder. A learner who forgets content is not thereby diagnosed with a memory disorder.

The Diagnostic Tutor should keep educational observations specific and know when a question belongs outside tuition.

Good diagnosis includes a boundary around the authority of the diagnosis.

27. A Diagnostic Session Should Produce a Better Next Action

A diagnosis that ends with a sophisticated description but no improved decision has limited value.

A useful diagnostic session should narrow the next move.

Before diagnosis	After useful diagnosis
“Weak in fractions”	Can compare fractions but cannot reconstruct equivalence when denominators differ
“Bad comprehension”	Finds evidence but answers the topic rather than the exact question demand
“Careless in Science”	Correct concept selection; loses causality when translating mechanism into written explanation
“Slow in algebra”	Method is stable; arithmetic retrieval is consuming most of the time
“Cannot remember”	Recognition is strong immediately; retrieval fails after delay without cues

Now teaching can become selective.

28. Repair One Mechanism, Then Ask It to Prove Itself

A diagnosis is a hypothesis about what is producing the difficulty.

The repair is therefore also a test.

If we believe the weak link is representation, improve representation and see whether downstream performance changes.

If we believe the weak link is retrieval, practise retrieval and see whether delayed access improves.

If we believe the weak link is vocabulary, clarify the critical language while holding the subject mechanism constant and see whether the learner can now reason correctly.

If performance does not change as predicted, the tutor should update the diagnosis rather than forcing the learner through more of the same intervention.

A good diagnosis is allowed to be wrong. What makes it good is that new evidence can correct it.

29. Reattempt Is Stronger Evidence Than Correction Alone

The tutor explains the error.

The student nods.

That is not yet enough.

The learner should reattempt.

Then the learner should attempt a changed item.

Then, where the skill matters, the tutor should return later.

This sequence turns a pleasant explanation into evidence about whether the repair actually entered the learner’s usable system.

It also aligns with the wider evidence around feedback and self-regulation: feedback matters when the learner can use the information to improve later work, not merely agree with it in the moment.

30. A Diagnostic Tutor Watches Prompt Count

Marks can improve while dependence remains unchanged.

That is why prompt count is useful.

Faith answers six questions correctly with six tutor prompts.

A week later she answers five correctly with one prompt.

The second performance may contain stronger evidence of improvement even though the raw score is slightly lower.

The Diagnostic Tutor therefore tracks support as part of the performance condition.

Useful questions include:

How many prompts were needed?
What kind of prompts?
At which step?
Could the prompt be reduced?
Did the learner begin generating the prompt internally?

31. Diagnose Across Time, Not Only Across Questions

Learning changes with time.

A concept can look stable immediately after teaching and become unavailable after a weekend.

A method can survive a day but disappear when another topic intervenes.

A vocabulary word can be recognised in a list but fail to appear during composition.

Diagnosis therefore needs temporal contrast.

Time point	What it can reveal
Immediate	Can the learner use the repaired idea while support is fresh?
Later in the session	Does the learner still need the same cue?
Next lesson	Was the learning retained?
Changed topic mix	Can the learner select the idea without chapter cues?
Examination return	Does the capability survive pressure and uncertainty?

A Diagnostic Tutor does not need to test every skill at every interval. The point is to use time when time can separate competing explanations.

32. Diagnostic Tutoring in a Three-Student Group

Small-group tuition creates a useful diagnostic environment when the tutor protects independent thinking.

Alicia, Beatrice and Ciara may work on the same question and fail in three different places.

Alicia misreads the condition.

Beatrice understands the condition but chooses an inefficient representation.

Ciara chooses the right method but makes a procedural error.

The shared task gives the tutor a controlled comparison. The learners are seeing a similar problem, but their routes diverge.

This is where small groups can offer something different from both one-to-one tutoring and large classes: enough closeness to see individual working, with enough variation to compare different learner routes.

The group should never become a diagnostic spectacle. Students do not need public labels. The tutor can use contrast quietly and teach shared mechanisms where useful.

33. Parents Need the Observation, Not the Drama

Parents naturally want a clear answer.

“What is wrong?”

A responsible Diagnostic Tutor may sometimes answer:

We have narrowed it, but we do not know yet.

That can be a high-quality answer when the evidence is genuinely incomplete.

A useful parent update has four parts:

Observation: what happened repeatedly?
Current hypothesis: what might explain it?
Evidence boundary: what do we still not know?
Next test or repair: what will we do to narrow or improve it?

This is calmer and more useful than turning one difficult week into a permanent identity.

Parents who want the wider decision map can use the Parents’ Guide to eduKate Sengkang.

34. Twelve Common Diagnostic Tutor Failure Modes

Failure 1: Diagnosing from the total score

The total says something happened. It rarely identifies exactly where the route broke.

Failure 2: Treating the first visible error as the first cause

The wrong final line may be downstream from reading, representation or method selection.

Failure 3: Choosing one explanation too early

“Careless” becomes a story that absorbs every later error.

Failure 4: Testing everything

Maximum testing is not maximum diagnosis. The best test is one that can change the decision.

Failure 5: Ignoring support conditions

Supported success and independent success are not interchangeable evidence.

Failure 6: Confusing confidence with competence

A confident answer may be wrong. A hesitant answer may be correct. Confidence is useful evidence only when interpreted alongside performance.

Failure 7: Confusing speed with understanding

Some learners are slow because they do not know. Others are slow because they are checking carefully. The observation must be decomposed.

Failure 8: Repairing the whole syllabus

A bounded problem should not automatically trigger a complete restart.

Failure 9: Never retesting the diagnosis

If the repair does not change the predicted behaviour, update the model.

Failure 10: Treating the child as the only variable

Task design, support, timing, marking and context also affect performance.

Failure 11: Diagnosing beyond educational authority

Tutors should not turn ordinary observations into medical or psychological diagnoses.

Failure 12: Making diagnosis permanent

The point is to repair and release, not to keep the learner permanently classified as a problem.

35. The Diagnostic Tutor’s Evidence Ladder

Not all evidence carries the same weight.

Evidence	What it can support
One wrong answer	A local error occurred
Repeated similar errors	A pattern may exist
Pattern across changed examples	The difficulty may belong to a deeper operation rather than one item
Success after a targeted cue	The cue identifies a potentially important missing step
Success after repair	The intervention may be working
Success with less support	More control has moved to the learner
Success after delay and variation	The repaired capability is becoming more stable and portable

The Diagnostic Tutor moves upward only when the evidence earns it.

36. A Six-Week Diagnostic Story: Denise and the Mathematics Problem That Was Not Where It Looked

Denise is in Secondary 2.

Her recent Mathematics paper shows repeated failure in algebra word problems. The obvious plan is an algebra revision programme.

Week 1: Bound the problem

Denise solves routine linear equations accurately. She simplifies expressions well. Her arithmetic is sufficiently fluent.

The problem narrows: she struggles mainly when relationships must be extracted from language.

Week 2: Test representation

The tutor gives the same relationship in three forms: sentence, table and diagram.

Denise succeeds with the table and diagram but misreads comparative phrases in sentences.

The hypothesis shifts from “algebra weakness” to “language-to-representation weakness inside algebraic contexts”.

Week 3: Repair the translation step

The tutor works on a small set of relationship phrases, asking Denise to convert each into a sketch, table or equation before solving anything.

The repair is narrow.

That is deliberate.

Week 4: Remove the cue

The tutor stops announcing that the task is about translation. Denise now receives mixed algebra questions.

She identifies the relationship independently in most of them.

Week 5: Change the surface

The contexts change from money and ages to distance, geometry and mixtures. Denise still reconstructs the relationships.

Week 6: Return after delay

The tutor returns to the skill after other topics have intervened. Denise remains mostly stable.

The original diagnosis has earned stronger support because a targeted repair improved the predicted mechanism and the improvement survived reduced cues, changed contexts and time.

This is what makes diagnosis different from storytelling.

37. The Diagnostic Tutor’s Session Template

Choose one real performance problem. Do not begin with the whole child.
Collect the best available artifact. Marked paper, exercise, writing sample, oral explanation or observed attempt.
Describe what happened. Avoid labels.
Find the boundaries. Where does the problem appear and disappear?
Decompose the task. What smaller operations are involved?
Keep two or three explanations alive. Do not force certainty.
Run one discriminating test. Change the condition that should separate the explanations.
Choose the earliest useful weak link. Not every weakness deserves immediate repair.
Apply the smallest appropriate intervention.
Reattempt. Same mechanism, fresh item.
Reduce support.
Return later if the skill matters.
Update the diagnosis.
Hand off to the correct tutor class when the job changes.

38. What Parents Should Bring to a Diagnostic Tuition Conversation

The most useful material is usually ordinary.

A recent marked school paper.
One or two pieces of normal homework.
Corrections showing what happened after feedback.
Teacher comments where available.
An example of work the learner finds unusually difficult.
An example the learner can do comfortably.
Information about what support was present.
A clear description of what changed recently, if anything.

Parents do not need to arrive with a diagnosis.

The purpose of the conversation is to improve the question before deciding the intervention.

39. The Full Tutor-Class Handoff

Class	Main job	When Class 3 hands over
0 · Homework Helper	Routine, organisation and completion support	Diagnosis shows the learner broadly has the capability and mainly needs structure
1 · Explainer	Make unclear ideas understandable	Diagnosis locates a genuine conceptual misunderstanding
2 · Drill Builder	Build fluency, speed and reliable execution	Diagnosis shows the method is sound but insufficiently stable
3 · Diagnostic Tutor	Find the cause before repair	Stays active while competing causes remain unresolved
4 · Route Designer	Sequence repair and development across time	Diagnosis reveals several interacting gaps requiring prioritisation
5 · Performance Coach	Prepare stable knowledge for examination output	Diagnosis shows ordinary capability is stable but pressure conditions remain weak
6 · Learning Architect	Integrate the whole learning system	Multiple layers interact and no single class can own the problem alone

The best Diagnostic Tutor does not try to keep every case in Class 3.

The best Diagnostic Tutor identifies the right next owner.

40. Frequently Asked Questions

Does every student need a Diagnostic Tutor?

No. Many learners have straightforward needs. A clear explanation or well-designed practice may be enough. Class 3 becomes valuable when performance is repeatedly weak, the cause is uncertain, or different interventions have not solved the problem.

Is diagnosis just giving the student a test?

No. A test provides evidence. Diagnosis is the reasoning process that uses evidence to distinguish plausible causes and choose a better next action.

Can one lesson diagnose everything?

Usually not. One lesson can narrow important problems, but some distinctions require repeated evidence, changed conditions or a return after delay. Good tutors keep uncertainty visible when the evidence is incomplete.

Why not just start practising and see what happens?

Sometimes that is efficient. But when the same difficulty keeps returning, more volume may rehearse the symptom rather than repair the cause. A small diagnostic check can prevent hours of low-value practice.

What makes a good diagnostic question?

It creates information that can change the decision. It may remove a cue, change a representation, simplify language, introduce a delay, compare timed and untimed conditions or ask the learner to explain why a tempting alternative is wrong.

How do we know the diagnosis was correct?

We rarely know with absolute certainty. Confidence rises when the diagnosis explains the observed pattern, predicts what should improve after targeted repair, and the predicted improvement survives fresh tasks, reduced support and appropriate return tests.

Can a strong student benefit from diagnosis?

Yes. High overall performance can hide bottlenecks in transfer, explanation, timing, method selection or advanced topic dependencies. Diagnosis is not a remedial label; it is a way to locate the next meaningful constraint.

What is the danger of a wrong diagnosis?

The learner may spend large amounts of time practising the wrong thing, receive a misleading label or become dependent on support that never reaches the true break. That is why diagnoses should remain testable and revisable.

When should the tutor stop diagnosing and start teaching?

As soon as the evidence is sufficient to justify a useful next action. Diagnosis is not an excuse for endless analysis. The purpose is better teaching.

41. Evidence and Further Reading

This handbook combines eduKate’s tutor-classification and first-weak-link architecture with established evidence themes around formative assessment, feedback, intervention, representations and learner independence. Useful starting points include:

42. Final Compression

A Diagnostic Tutor does not make tutoring complicated for the sake of complexity.

The role exists to stop complexity from being hidden inside vague labels.

The student gets the question wrong.

That is the beginning of the investigation, not the end.

What did the learner actually do?

Where did the route first change?

What else could explain the same symptom?

What small test would separate those explanations?

What is the earliest useful weak link?

What is the smallest appropriate repair?

Did the repair work on a fresh problem?

Did it survive with less help?

Did it remain available later?

And if the answer changes, are we willing to change the diagnosis?

The goal is not to become better at explaining why a learner is weak. The goal is to become better at locating what can be repaired so the learner can become stronger.

That is the discipline of the Diagnostic Tutor.

That is Tutor Handbook Volume 0004.

Next in the series: The Tutor Handbook Vol No.0005 | Route Designer — Build the Learning Path in the Right Order.

Small Group Tutorials

The Tutor Handbook Vol No.0004 | Diagnostic Tutor — Find the First Weak Link Before Adding More Work