G2 SEC examination performance requires two kinds of correctness at once. A sentence, calculation or observation can be locally correct and still fail because it contradicts the wider passage, question, model, dataset, instruction or real-world constraint. This is a local–global consistency problem.
This one-hundred-and-eighth Learner’s Guide develops local–global consistency across K200 English, K210 Mathematics and K223–K225 Science. It builds on Vol 0104 Representation Handoffs and Vol 0092 Instruction Hierarchy.
The two-level check
First ask whether the local step is valid on its own. Then ask whether it remains compatible with the larger structure: source, chronology, units, domain, sample space, dataset, control, scope, instruction or task. A correct local move cannot override a violated global constraint.
1. Local answer versus whole-paper instruction
Local state: One response is correct locally but violates a paper-wide rule.
Global check: The global rule remains active; repair only the interface required by it.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
2. Subpart versus question stem
Local state: Part (b) is answered correctly but ignores a condition stated above all subparts.
Global check: Carry inherited conditions downward unless explicitly replaced.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
3. Calculation versus context
Local state: Arithmetic is correct but the answer is impossible for people, capacity or time.
Global check: Context is a global constraint on the local number.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
4. Sentence versus paragraph
Local state: One English sentence sounds plausible but contradicts the paragraph’s final qualification.
Global check: Use the local evidence window plus paragraph stance.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
5. Paragraph versus whole text
Local state: A paragraph is positive while the conclusion is cautious.
Global check: Do not use one local tone as the whole-text attitude.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
6. Quotation versus source
Local state: A phrase supports the idea but belongs to another speaker.
Global check: Source ownership overrides semantic fit.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
7. Pronoun versus discourse
Local state: Nearest noun seems possible but wider discourse makes another referent coherent.
Global check: Use grammar plus global coherence.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
8. Word meaning versus passage
Local state: Dictionary meaning fits locally but breaks the surrounding argument.
Global check: Contextual meaning must cohere with the passage.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
9. Inference versus chronology
Local state: A motive fits the action but depends on information learned later.
Global check: Global timeline rejects the local inference.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
10. Cause versus whole mechanism
Local state: One causal link is valid but another required link is missing.
Global check: Do not let a local truth stand in for the complete explanation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
11. Summary point versus source
Local state: A compressed point is accurate but duplicates another point.
Global check: Global summary coverage requires uniqueness.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
12. Writing sentence versus task
Local state: A polished sentence does not fulfil any prompt requirement.
Global check: Task-level purpose determines whether it belongs.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
13. Writing paragraph versus composition
Local state: A paragraph is strong alone but derails narrative/argument progression.
Global check: Whole-text function matters.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
14. Oral example versus position
Local state: Example is vivid but contradicts stated recommendation.
Global check: Align example with global stance.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
15. Math step versus equation
Local state: One algebraic manipulation is legal but begins from a wrong model.
Global check: Local algebra cannot rescue global representation error.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
16. Math answer versus units
Local state: Number is plausible but quantity type is wrong.
Global check: Dimensional consistency rejects it.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
17. Math root versus domain
Local state: Root satisfies transformed equation but violates original domain.
Global check: Original problem is the global authority.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
18. Math branch versus case set
Local state: One case is solved correctly but another valid case is omitted.
Global check: Global solution set requires exhaustion.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
19. Math estimate versus exact
Local state: Exact result is outside a justified bound.
Global check: Recheck local execution.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
20. Math graph point versus trend
Local state: One point is read correctly but the claimed trend uses wrong interval.
Global check: Global requested interval controls interpretation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
21. Math mean versus data
Local state: Computed mean is outside min–max range.
Global check: Dataset-level invariant rejects local arithmetic.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
22. Math probability versus universe
Local state: Fraction arithmetic is correct but event universe is wrong.
Global check: Sample-space ownership is global.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
23. Math rate versus denominator
Local state: Division is correct but denominator answers another rate.
Global check: Question-level quantity identity controls.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
24. Math rounding versus constraint
Local state: Rounded value is numerically nearest but violates minimum requirement.
Global check: Feasibility outranks ordinary rounding.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
25. Science observation versus conclusion
Local state: Observation is correct but conclusion exceeds tested range.
Global check: Scope must match experiment.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
26. Science treatment result versus control
Local state: Treatment changes but control changes similarly.
Global check: Global comparison weakens treatment attribution.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
27. Science local measurement versus system
Local state: One sensor reading is accurate but unrepresentative.
Global check: Spatial scale limits generalisation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
28. Science one trial versus pattern
Local state: One result is valid but repeats vary.
Global check: Dataset-level evidence controls confidence.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
29. Science mechanism versus data
Local state: Mechanism is scientifically plausible but measured trend opposes it.
Global check: Live evidence requires revision or boundary.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
30. Science model versus assumption
Local state: Prediction is correct only under an assumption that failed.
Global check: Model validity is conditional.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
31. Science anomaly versus trend
Local state: One point conflicts with otherwise stable pattern.
Global check: Preserve both trend and anomaly.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
32. Science negative result versus sensitivity
Local state: No signal appears but method is weak.
Global check: Measurement capability limits global absence claim.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
33. Science positive result versus specificity
Local state: Signal appears but another cause can produce it.
Global check: Test validity limits conclusion.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
34. Science subgroup versus population
Local state: One group responds differently.
Global check: Do not average away the boundary.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
35. Science short-term versus long-term
Local state: Immediate effect differs later.
Global check: Time horizon is part of claim.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
36. Instruction versus habit
Local state: Familiar question type suggests one method but stem asks another.
Global check: Live instruction overrides template.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
37. Current question versus earlier question
Local state: Same topic appears with different conditions.
Global check: Do not carry assumptions across question boundaries.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
38. Current section versus previous section
Local state: Response mode changes.
Global check: Reset section-specific rules.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
39. MCQ option versus stem negation
Local state: Option is true but stem asks which is false.
Global check: Global selection criterion controls.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
40. Evidence detail versus command word
Local state: Detail is relevant but question asks explanation.
Global check: Add mechanism rather than more detail.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
41. Answer length versus mark function
Local state: Long response contains only one distinct point.
Global check: Count functions, not words.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
42. Confidence versus contradiction
Local state: Learner feels sure but independent check fails.
Global check: Evidence outranks confidence.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
43. Uncertainty versus stable branches
Local state: One unknown remains but other work is secure.
Global check: Quarantine uncertainty locally.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
44. Repair versus independent work
Local state: One intermediate changes.
Global check: Propagate only through descendants.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
45. Checkpoint versus full reconstruction
Local state: Return note preserves next action.
Global check: Resume from compressed state, not from zero.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
46. Representation versus invariant
Local state: Equation and graph look different.
Global check: Check whether they preserve the same relation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
47. Second representation versus first
Local state: Two forms disagree.
Global check: Locate conversion error rather than choosing favourite.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
48. Final answer versus working
Local state: Working is correct but copied result differs.
Global check: Transfer integrity is part of global consistency.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
49. Label versus value
Local state: Value is correct but attached to wrong row/source.
Global check: Ownership matters.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
50. Unit versus numerical magnitude
Local state: Conversion changes number.
Global check: Physical quantity should remain invariant.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
51. Quantifier versus evidence
Local state: Evidence covers most, answer says all.
Global check: Scope mismatch is a global inconsistency.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
52. Modality versus evidence
Local state: Source says may, answer says will.
Global check: Certainty must not exceed support.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
53. Exception versus rule
Local state: General pattern holds except one condition.
Global check: Store boundary with rule.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
54. Presupposition versus assertion
Local state: Sentence assumes background not explicitly proved.
Global check: Keep hidden assumption source-bound.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
55. Causal connector versus sequence
Local state: Events occur in order.
Global check: Chronology alone does not justify because.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
56. Comparison versus criterion
Local state: A is called better without named dimension.
Global check: Global comparison requires criterion.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
57. Percentage versus base
Local state: Percentage looks right but uses wrong reference.
Global check: Denominator defines meaning.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
58. Difference versus ratio
Local state: Absolute gap is correct but question asks relative change.
Global check: Quantity type controls operation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
59. Graph scale versus visual slope
Local state: Line looks steep because axis scale changes.
Global check: Numeric scale controls interpretation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
60. Table total versus category
Local state: Category value is correct but exceeds stated total.
Global check: Whole–part invariant rejects it.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
61. Geometry diagram versus theorem
Local state: Drawing suggests equality but none is stated.
Global check: Formal relation outranks appearance.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
62. Probability path versus total
Local state: One path is correct but paths overlap.
Global check: Global counting must avoid duplication.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
63. Science variable versus method
Local state: Named variable is correct but not actually manipulated.
Global check: Procedure defines role.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
64. Science control versus claim
Local state: Control removes one alternative only.
Global check: Do not claim all confounding eliminated.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
65. Science interaction versus main effect
Local state: A matters at one B level but not another.
Global check: Conditional effect is more coherent than universal statement.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
66. Science scale versus mechanism
Local state: Cell-level process is correct but organism result needs bridge.
Global check: Cross-scale coherence requires aggregation.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
67. Paper timing versus one hard item
Local state: Question could be solved with enough time.
Global check: Whole-paper mark opportunity limits local investment.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
68. Review versus completion
Local state: One answer can be polished further.
Global check: Global unfinished work may have higher recoverability.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
69. Choice question versus both options
Local state: Both are partially attempted.
Global check: Paper-level choice rule requires one final route.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
70. Submission versus scratch state
Local state: Correct idea remains only in rough work.
Global check: Final response surface must contain it.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
71. Local-global final rule
Local state: Every answer must be locally valid and globally compatible.
Global check: When levels disagree, locate the higher constraint that the local step violated.
For practice, create one answer that passes the local check but fails the global one. Then repair only the level that is inconsistent. This trains learners to diagnose why a plausible answer is still wrong instead of simply redoing everything.
Consistency is not conformity
Global consistency does not mean every detail must point in the same direction. A text can contain a deliberate exception; data can contain an anomaly; a system can contain opposing mechanisms. The requirement is that the final interpretation accounts for those differences rather than silently contradicting them.
Links
Use the Examination Craft hub, Vol 0100 Checkpoint Compression, the relevant English, Mathematics and Science hubs, and the PSLE Learner’s Guide.
Official-source discipline
For the 2027 SEC G2 school-candidate framework, use the current SEAB G2 syllabus directory and linked subject syllabuses. Local–global consistency is an eduKateSengkang training framework, not an SEAB instruction.
Final rule
Do not ask only whether a step is correct. Ask whether it can coexist with everything else that must also be true. Examination reliability comes from local validity inside global constraints.
