Diagnostic teaching is the process of locating the earliest useful reason a learner’s performance breaks down, then changing instruction at that point instead of treating every wrong answer as the same problem. A student may fail a mathematics question because of a missing prerequisite, a misread command, a representation error, the wrong method, an execution slip or a failure to check. A writer may produce a weak paragraph because the evidence is missing, the claim is unclear, the structure is wrong for the purpose or the language cannot yet carry the idea. For readers searching for diagnostic teaching, learning gaps, misconceptions, responsive teaching, formative assessment or how to find why a student is struggling, the central mechanism is observe → locate → discriminate → repair → retest → transfer.
Diagnosis matters because visible failure is downstream. By the time an answer is wrong, several earlier decisions may already have gone wrong. Correcting only the final answer can make the page look better while leaving the cause untouched. Effective diagnostic teaching works backwards through the dependency chain until it finds the narrowest bottleneck that explains enough of the failure to guide a useful next action.
This world-facing guide develops diagnosis across mathematics, science, reading, writing, creative work, discussion, homework, study, motivation, scaffolding and assessment. Adrian, Jo, Ben, Aisha, Ryan, Mira, Clara and Ethan appear only as fictional eduKateSG residents in invented teaching scenarios. They are not testimonials. The examples are teaching constructions, not clinical or psychological diagnoses. Persistent learning, developmental, health or access concerns require the appropriate professional and institutional processes.
Quick Read: find the first wrong turn, not only the wrong destination
Ben writes 36 m as the area of a rectangle. Adrian could correct the unit to 36 m². Instead he asks, “What quantity did you calculate?” Ben answers, “How far around it.” The visible unit error is not the first problem. Ben has selected perimeter while believing he selected area. Another ten area calculations will not repair the selection rule unless the distinction becomes explicit.
Adrian places two nearly identical questions side by side: one asks for fencing, the other for floor covering. Ben identifies what is being measured before calculating. A later unfamiliar problem removes the familiar rectangle and asks about edging versus surface material. Diagnosis has become repair only when the changed task shows that the distinction travels.
For the broader architecture, begin with How Education Works. Feedback Loops owns the response cycle, Scaffolding owns temporary support, and Prerequisite Gaps owns dependency repair. This article owns the locating mechanism: how teachers decide where to intervene before they intervene.
1. The wrong answer is evidence, not a diagnosis
Two students can write the same wrong answer for different reasons. One misunderstood the concept. Another knew the concept but copied a number incorrectly. A third selected the wrong procedure because the problem resembled yesterday’s example. Treating all three with the same reteaching wastes information.
The first diagnostic job is to preserve the learner’s process long enough to inspect it. Ask for the representation, explanation, intermediate step or source used. Do not overwrite the evidence before understanding what it reveals.
Diagnosis is useful only when it changes the next teaching action. A label that sounds precise but offers no repair route is not enough.
2. Observe before explaining
Let the learner attempt enough of the task to reveal the process. Immediate rescue can erase evidence needed for diagnosis.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
3. Locate the first break
Trace backwards from the wrong answer to the earliest decision that no longer follows. Earlier repairs often resolve downstream errors.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
4. Use discriminating questions
Ask a question whose possible answers separate competing explanations. What quantity are you finding separates selection from calculation.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
5. Preserve intermediate work
Cross-outs, diagrams and partial steps reveal reasoning. Do not value a clean page more than diagnostic evidence.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
6. Knowledge gap
If the learner cannot recall or explain required knowledge, teach or retrieve it before blaming strategy.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
7. Prerequisite gap
A current topic may fail because an earlier dependency is missing. Trace the chain before adding current-topic practice.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
8. Misconception
A misconception is a systematic model that produces predictable errors. Test it across contrasting cases.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
9. Slip
A slip occurs despite otherwise secure knowledge. Look for consistency across nearby tasks before launching full reteaching.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
10. Reading error
The learner may know the subject but misread the command, qualifier or data. Separate reading from subject knowledge.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
11. Representation error
A diagram, equation, graph or table may encode the problem incorrectly before calculation begins.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
12. Method selection
Students can execute several methods yet choose the wrong one. Mixed and contrasting problems reveal selection rules.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
13. Execution error
The method is appropriate but a step is performed incorrectly. Target the step rather than reteaching the entire topic.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
14. Checking failure
The learner reaches an implausible answer but does not notice. Teach estimation, units, substitution or source checking as part of the method.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
15. Language bottleneck
A learner may understand the concept but lack language to express it. Test understanding through another appropriate mode.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
16. Attention lapse
One inconsistent error may reflect missed information rather than a stable gap. Recheck before inferring a misconception.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
17. Task ambiguity
Sometimes the problem is the question. If capable learners interpret it in several reasonable ways, revise the task.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
18. Instruction gap
If many learners fail at the same hidden step, inspect what was taught and modelled. Class-wide error can reveal an instructional omission.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
19. Curriculum gap
A required prerequisite may never have been securely taught in the sequence. Diagnosis can point beyond the individual learner.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
20. Assessment mismatch
A task may measure language, speed or formatting more than the intended construct. Diagnose the assessment before diagnosing the student.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
21. Support dependence
A learner succeeds only with a prompt, model or checklist. Identify which decision the support is carrying.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
22. Prompt dependence
Delay or alter the familiar prompt. If performance collapses, teach the learner to generate the cue internally.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
23. Model dependence
Change the surface of a worked example. Copying success that disappears under variation indicates narrow learning.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
24. Peer dependence
Ask for an individual reconstruction after collaboration. Group fluency can hide individual gaps.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
25. Teacher dependence
Increase distance gradually. Waiting for confirmation after every step suggests monitoring remains externally controlled.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
26. Motivation versus capability
A blank page can reflect low value, low expectancy, confusion or missing knowledge. Do not diagnose motivation from non-completion alone.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
27. Start-cost diagnosis
Ask the learner to show the first action. If they cannot, ambiguity or planning may be the bottleneck.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
28. Effort diagnosis
More effort does not repair the wrong method. Identify whether strategy or knowledge must change.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
29. Feedback diagnosis
Feedback should match the located gap rather than the most visible symptom.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
30. Practice diagnosis
Repeated errors show that the practice loop needs interruption and repair.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
31. Homework diagnosis
Missing homework can reveal readiness, access, planning or motivation problems.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
32. Study diagnosis
Long study time with weak retrieval suggests method mismatch.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
33. Discussion diagnosis
Classroom talk reveals misconceptions when questions discriminate between explanations.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
34. Collaborative diagnosis
Compare private answers before and after group work. This shows what the group changed.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
35. Modelling diagnosis
Student errors can reveal hidden expert steps omitted from a demonstration.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
36. Mathematics diagnosis
Separate concept, representation, method selection, execution, units and checking.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
37. Science diagnosis
Separate factual knowledge, variable reasoning, observation, inference, mechanism and evidence scope.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
38. History diagnosis
Separate chronology, source comprehension, provenance, contextualisation, evidence selection and interpretation.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
39. Geography diagnosis
Separate scale, spatial reading, data interpretation, comparison and causal inference.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
40. Reading diagnosis
A comprehension error may begin in decoding, vocabulary, syntax, background knowledge, inference or answer scope.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
41. Vocabulary diagnosis
A learner may know a definition but not usage, collocation, register or near-synonym distinctions.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
42. Writing diagnosis
Separate idea, evidence, organisation, sentence construction, vocabulary, transcription and revision.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
43. Creative-writing diagnosis
Separate premise, conflict, viewpoint, sequence, detail selection, dialogue and ending.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
44. Oral diagnosis
Separate knowledge, planning, language retrieval, listening, organisation and delivery.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
45. Exam diagnosis
Classify errors by knowledge, interpretation, selection, execution, timing and checking.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
46. Timed-work diagnosis
Compare untimed and timed performance. If both are weak, speed may not be the first target.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
47. Data diagnosis
Inspect denominator, units, scale, sample and comparison before interpreting conclusions.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
48. Graph diagnosis
Ask what each axis represents and why the graph type fits.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
49. Source diagnosis
Ask what the source states, what the learner infers and what remains unknown.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
50. Argument diagnosis
Locate whether the weakness lies in claim, evidence, warrant, counterargument or qualification.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
51. Explanation diagnosis
A learner may describe what happens without explaining why. Ask for the relationship or mechanism.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
52. Creative-choice diagnosis
A writer may have many ideas but no selection criterion. Ask what effect the scene is intended to produce.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
53. Transfer diagnosis
Success on familiar items but failure on changed tasks indicates narrow encoding, cue dependence or selection problems.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
54. Retention diagnosis
Success today and failure after delay suggests retrieval strength is fragile. Add spaced return.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
55. Fluency diagnosis
Slow but accurate performance differs from fast inaccurate performance. Speed is a separate dimension.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
56. Automaticity diagnosis
A basic process may consume so much attention that higher-level reasoning collapses. Practise the component while preserving understanding.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
57. Confidence diagnosis
Confidence can be high or low independently of accuracy. Compare predictions with outcomes.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
58. Calibration
Ask learners to rate uncertainty before checking. Compare confidence with accuracy.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
59. Help-seeking
A vague I do not understand can be narrowed: where does understanding stop?
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
60. Error patterns
One error is evidence; repeated related errors create a pattern. Use enough cases to distinguish gaps from noise.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
61. Contrast cases
Present examples that differ in one important feature. Contrast reveals the learner’s selection rule.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
62. Near misses
A nearly correct response can reveal a single missing relation. Preserve what works while repairing the narrow gap.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
63. Multiple-choice distractors
Well-designed distractors can represent plausible misconceptions. The selected option is useful when the distractor has a diagnostic rationale.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
64. Open responses
Open work reveals more process but is slower to inspect. Use it when the reasoning path matters.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
65. Mini-whiteboards
Simultaneous responses reveal class-wide patterns quickly. Follow the pattern with probing.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
66. Exit tickets
A short end-of-lesson task can identify what remains fragile and inform the next lesson.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
67. Low-stakes quizzes
Quizzes can combine retrieval and diagnosis. Use item-level patterns rather than only total scores.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
68. Interviews
A short conversation can distinguish knowledge from expression and reveal why a method was selected.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
69. Think-aloud diagnosis
Ask a learner to verbalise selected decisions during a task without assuming every internal process can be verbalised.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
70. Observation
Watch where the learner pauses, checks, skips or seeks confirmation. Behaviour is evidence, not a complete explanation.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
71. Work samples
Compare work across time and conditions. A single polished product can hide extensive support.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
72. Baseline
Collect an initial sample before an intervention when feasible.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
73. Retest
After repair, use a matched item to see whether the specific gap changed.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
74. Delayed retest
Return later without the correction visible. This tests whether the repair survived.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
75. Transfer retest
Change context or representation. This tests whether the learner acquired a portable rule.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
76. Repair hierarchy
Repair the earliest high-leverage gap first. Do not correct downstream symptoms while leaving the dependency broken.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
77. Micro-teaching
Teach the narrow missing distinction, then immediately retest.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
78. Reteaching
Broader reteaching is appropriate when evidence shows a wider conceptual gap, not automatically after every error.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
79. Alternative representation
A diagram, concrete model or verbal explanation can reveal whether the difficulty is conceptual or representational.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
80. Worked-example repair
Return to a model when the learner lacks a procedure, then fade again after successful reconstruction.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
81. Scaffold repair
Add a temporary cue when the learner knows the content but cannot manage the process independently.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
82. Vocabulary repair
Clarify a word when language blocks access to the actual subject target.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
83. Reading repair
Chunk the text, clarify syntax or build missing background knowledge according to the barrier.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
84. Motivation repair
If capability is adequate but initiation remains low, inspect value, expectancy and cost.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
85. Access repair
If format creates an irrelevant barrier, use agreed accessible alternatives.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
86. Multilingual learners
Test concept understanding separately from language production where appropriate.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
87. Accessibility
Distinguish temporary instructional supports from ongoing accommodations.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
88. Parents
Parents can report where homework stops and what support was used. Avoid repairing the page so completely that evidence disappears.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
89. Tutors
Tutors should preserve the original attempt, identify the first bottleneck and test the repair independently.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
90. Teacher teams
Compare interpretations of anonymised work. Disagreement can reveal unclear criteria or hidden assumptions.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
91. Curriculum teams
Aggregate recurring bottlenecks to identify sequencing gaps, duplicated effort or missing prerequisites.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
92. Assessment teams
Inspect whether items produce interpretable evidence for the intended claim.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
93. AI-assisted diagnosis
Where permitted, AI may help classify patterns or generate contrast items, but outputs require human verification and approved data handling.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
94. Privacy
Collect only diagnostic information needed for legitimate educational purposes.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
95. Labels
Use conditional descriptions rather than permanent identities. Needs a prompt to distinguish area from perimeter is more actionable than weak at maths.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
96. Clinical boundaries
Teachers can diagnose instructional gaps; they should not infer medical, neurodevelopmental or mental-health conditions from ordinary classroom errors.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
97. Referral
When persistent concerns exceed ordinary instructional diagnosis, follow established school and professional referral processes.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
98. A diagnostic conference
Ask: What were you trying to do? Where did you become uncertain? What did you try next? What would distinguish this case from the other one?
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
99. A six-week implementation
Week one preserves process evidence. Week two improves discriminating questions. Week three links diagnoses to micro-repairs. Week four adds delayed retests. Week five adds transfer. Week six audits precision.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
100. Measuring diagnostic quality
A diagnosis is useful when the predicted repair improves the targeted performance. Accuracy of the label is tested by the return.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
101. Failure mode: correcting too early
The teacher supplies the answer before locating the gap. Preserve enough process to diagnose.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
102. Failure mode: one error, big label
A single mistake becomes a claim about ability. Gather pattern evidence.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
103. Failure mode: reteach everything
Broad reteaching follows a narrow slip. Use the smallest repair that fits.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
104. Failure mode: blame motivation
Confusion is called laziness. Check capability and task clarity first.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
105. Failure mode: blame ability
A missing prerequisite becomes a fixed trait. Describe the dependency and repair it.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
106. Failure mode: score-only diagnosis
A total score hides which mechanisms failed. Analyse item and process patterns.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
107. Failure mode: no retest
The teacher diagnoses and reteaches but never checks whether the repair worked.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
108. Failure mode: same retest
The learner memorises the correction. Change surface features.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
109. Failure mode: diagnostic overload
Teachers collect huge datasets they cannot act on. Gather the smallest evidence that changes instruction.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
110. Failure mode: hidden support
Performance is interpreted as independent although prompts were supplied. Record conditions.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
111. Research base
Diagnostic assessment, formative assessment, misconception research and responsive teaching provide overlapping foundations. Their value lies in adapting teaching, not elaborate labels.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
112. The final diagnosis
The best diagnosis is temporary and actionable. It tells us where learning currently breaks, what to try next and what evidence would make us change our mind.
In the fictional eduKateSG classroom, Adrian asks one discriminating question before offering a correction. Jo preserves the learner’s original attempt and notes the support conditions. The next action targets the earliest useful bottleneck rather than every visible imperfection. This keeps diagnosis tied to teaching rather than turning it into a label.
After the repair, the learner completes a matched item and later a changed one. If the predicted error disappears, the diagnosis gains support. If it remains, the team reopens the hypothesis. A good diagnostic system is willing to be wrong quickly and revise from evidence.
106. Thirty diagnostic laboratories
Laboratory 1: Area or perimeter
For area or perimeter, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest area or perimeter immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 2: Equivalent fractions
For equivalent fractions, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest equivalent fractions immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 3: Algebra sign error
For algebra sign error, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest algebra sign error immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 4: Scientific variables
For scientific variables, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest scientific variables immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 5: Causal explanation
For causal explanation, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest causal explanation immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 6: Historical chronology
For historical chronology, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest historical chronology immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 7: Source reliability
For source reliability, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest source reliability immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 8: Map scale
For map scale, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest map scale immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 9: Graph axes
For graph axes, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest graph axes immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 10: Reading vocabulary
For reading vocabulary, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest reading vocabulary immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 11: Reading inference
For reading inference, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest reading inference immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 12: Answer scope
For answer scope, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest answer scope immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 13: Vocabulary near-synonyms
For vocabulary near-synonyms, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest vocabulary near-synonyms immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 14: Essay claim
For essay claim, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest essay claim immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 15: Evidence integration
For evidence integration, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest evidence integration immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 16: Paragraph organisation
For paragraph organisation, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest paragraph organisation immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 17: Creative conflict
For creative conflict, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest creative conflict immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 18: Viewpoint
For viewpoint, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest viewpoint immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 19: Dialogue
For dialogue, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest dialogue immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 20: Story ending
For story ending, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest story ending immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 21: Oral response
For oral response, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest oral response immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 22: Exam timing
For exam timing, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest exam timing immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 23: Homework start
For homework start, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest homework start immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 24: Study method
For study method, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest study method immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 25: Group dependence
For group dependence, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest group dependence immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 26: Scaffold dependence
For scaffold dependence, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest scaffold dependence immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 27: Transfer failure
For transfer failure, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest transfer failure immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 28: Retention failure
For retention failure, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest retention failure immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 29: Confidence mismatch
For confidence mismatch, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest confidence mismatch immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
Laboratory 30: Unfamiliar final task
For unfamiliar final task, begin with an attempt that preserves process. Do not tell the learner which category of error you suspect. Ask them to show or explain the first decision. Identify two plausible causes and choose one question or contrast case that would make those causes produce different responses. Diagnosis improves when evidence can discriminate between hypotheses.
Once the bottleneck is located, repair only that relationship first. If the learner lacks knowledge, teach it. If method selection is wrong, contrast cases. If execution is unstable, use matched practice. If monitoring is missing, add a checking routine. If language blocks expression, support the language while preserving the concept. The repair should follow the diagnosis rather than a favourite intervention.
Retest unfamiliar final task immediately with a related item, then later with a changed surface. The first retest asks whether the learner can use the repair; the later test asks whether the learning survived. Record the support present. A correct response after a leading prompt and a correct independent response support different claims.
If the predicted improvement does not appear, reopen the diagnosis. Do not protect the original explanation because time was invested in it. The learner’s return is new evidence. Diagnostic teaching becomes rigorous when teachers are willing to update their own model of the problem.
107. Original story: The First Wrong Line
Ben’s answer was wrong by twelve.
That was what he noticed first. Adrian noticed the second line.
“Tell me what this number means,” Adrian said.
“The total.”
“Total of what?”
Ben looked at the diagram. “The outside.”
The question asked how much tile covered the floor.
“So my multiplication is wrong?” Ben asked.
“Your multiplication is correct.”
Ben frowned. Correct mathematics inside a wrong answer felt worse than an ordinary mistake.
Adrian drew two tiny rooms. Under one he wrote skirting board. Under the other he wrote tiles. “No calculation. What quantity does each job need?”
Ben pointed around the first rectangle and across the second.
“Now name them.”
“Perimeter. Area.”
The next question used a garden instead of a room. Ben underlined cover the ground before writing any numbers.
Adrian left the original wrong answer on the page. The useful part was not the correction at the bottom. It was the place where the reasoning first turned.
108. Reading the story diagnostically
Ben’s arithmetic is not the bottleneck. The first wrong line represents the wrong quantity. If Adrian had assigned multiplication practice, Ben could become faster at executing the wrong model. The contrast between skirting and tiles isolates the distinction without requiring a full reteaching sequence.
The garden question changes the surface while preserving the area-perimeter decision. Ben’s underlining is a self-generated cue. The later response provides more relevant evidence of repair than correcting the original number alone.
109. Diagnostic teaching and creative writing
A weak story should not receive a generic prescription to add description. First locate what prevents the story from working. If the character has no meaningful goal, more sensory detail may beautify a static scene. If the viewpoint shifts accidentally, vocabulary is not the first repair. If the ending feels unearned, inspect the earlier chain of choices.
Use a discriminating question: “What changes because the character makes this choice?” If the writer cannot answer, conflict or causality may be the bottleneck. “What can the viewpoint character know here?” diagnoses a different issue. “Which detail should the reader remember at the end?” targets selection and motif. Creative diagnosis preserves artistic choice while making craft problems actionable.
After repair, ask for a new short scene rather than endless polishing of the same story. Transfer reveals whether the writer learned a craft decision or merely followed feedback on one passage.
110. Diagnostic teaching and learner dignity
Diagnosis should reduce unnecessary judgement. “You are weak at comprehension” collapses many possible mechanisms into an identity. “You can locate explicit evidence, but this answer becomes too broad when the question requires an inference limited to paragraph four” describes a current capability and a repairable boundary.
Precision also protects high expectations. Teachers do not need to lower the target simply because one dependency is weak. They can repair the dependency and return to the ambitious task. A learner is more than the bottleneck currently limiting performance.
Where persistent concerns suggest needs beyond ordinary teaching, use the school’s established support and referral processes. Instructional diagnosis should not become amateur clinical diagnosis.
111. Research and professional reading
Diagnostic teaching draws on formative assessment, assessment-for-learning, misconception research and responsive instruction. Black and Wiliam’s work on formative assessment emphasises using evidence of learning to adapt teaching. Research on misconceptions across subjects shows why wrong answers can reflect coherent but incorrect models rather than random failure. Cognitive diagnostic approaches in assessment pursue related questions at a more formal measurement level.
The classroom framework here is intentionally practical and non-clinical. It does not claim that every error can be assigned one cause with certainty. Its standard is usefulness: a working hypothesis should predict a repair, and the learner’s subsequent performance should be allowed to confirm, refine or reject that hypothesis.
112. A six-week implementation
Week one preserves intermediate work and samples common error patterns. Week two improves discriminating questions. Week three links each diagnosis to a small repair. Week four adds immediate and delayed retests. Week five changes representations and contexts. Week six audits whether teacher comments have become more conditional, actionable and efficient.
This sequence should reduce work as well as improve it. If teachers collect diagnostic information they cannot use, simplify. One well-chosen question that changes instruction is more valuable than a dashboard full of unacted data.
113. Connected routes through eduKate
Use Prerequisite Gaps when the bottleneck lies earlier in the dependency chain, Feedback Loops for the response, Scaffolding for temporary support, and Practice for repair and retesting.
114. The final diagnostic rule
Do not ask only, “What did the learner get wrong?” Ask, “What is the earliest useful place where their model of the task diverges from the one the task requires?” That question changes correction into teaching.
The answer should remain provisional. A diagnostic hypothesis earns confidence when the repair it predicts improves performance and survives a changed task. When it does not, update.
Education becomes more humane and more rigorous at the same time when failure is treated neither as a permanent label nor as something to ignore, but as evidence that can help us find the next teachable point.