Series route: This guide belongs to the complete PSLE Learner’s Guide volume directory. The original PSLE Learning Guide supplies the wider English, Mathematics and Science route, while PSLE English, PSLE Mathematics, PSLE Science provide verified subject-specific continuation.
Find your route through this guide
This volume solves one precise learner problem: the learner gathers more evidence that both rival explanations already predict, then calls the result decisive even though the test could not distinguish between them. The goal is durable control, not a remembered slogan. A Primary 6 learner should be able to identify the trigger, explain the mechanism, perform the decision under time pressure and reject an attractive answer that exceeds the evidence.
- Start with the exact mistake and the evidence boundary.
- Build the mechanism before memorising a shortcut.
- Study worked contrasts and name why the near-miss fails.
- Repair misconceptions with one targeted check each.
- Attempt graded independent practice before reading the explanations.
- Return after a delay and prove that the method transfers.
- Use the parent and tutor notes after the learner has attempted the work.
The problem beneath the mistake
Evidence is decisive between two explanations only when the explanations predict different outcomes for the test.
If both explanations predict the same result, observing that result may support both and eliminate neither.
In English comprehension, two interpretations may both predict that a character leaves. A decisive detail must reveal why the character leaves or which belief guided the action.
In Mathematics, two candidate rules may agree for the first three input values. A useful test input is one where their outputs diverge.
In Science, warmth and light may both change when a lamp moves closer to a plant. A comparison must control one factor if it is meant to identify the other.
The process begins before collecting evidence: state the rival explanations clearly enough to derive a prediction from each.
A fair test controls irrelevant differences, but fairness alone does not guarantee discrimination. Two fair setups can still produce a result expected under both explanations.
A null result also needs interpretation. If neither predicted outcome appears, the apparatus, assumption or set of alternatives may be incomplete.
Sensitivity matters as well as fairness. If two explanations predict a small difference but the measurement can detect only a large one, a recorded tie does not genuinely show that the predictions were equal. The learner should ask whether the instrument, sample or reading method could reveal the expected separation.
Decisive does not mean dramatic. A quiet grammatical clue, one carefully chosen input or a controlled observation may carry more logical weight than a large collection of vivid but non-discriminating details. The test earns strength from its relationship to the rivals, not from its size.
One discriminating result can reject a candidate under the tested conditions, but it should not be inflated into a universal claim beyond the domain.
The reliable routine is name rivals, find a point of disagreement, predict outcomes before testing, control other factors, observe, compare and bound the conclusion.
Core decision: A strong test separates rival explanations by making them predict different observable results. State predictions in advance, control other causes and interpret the outcome only within the tested conditions.
Evidence gains meaning through contrast. Collecting more observations is not automatically stronger than choosing one observation that bears differently on the competing claims.
Mechanism: why the rule works
1. State rival explanations
Write each candidate as a complete claim.
Vague rivals cannot generate testable predictions.
Diagnostic check. What exactly differs between the claims?
2. Find disagreement
Identify a case, input or condition where predictions diverge.
Agreement cases cannot discriminate.
Diagnostic check. Where would the rivals give different answers?
3. Predict in advance
Record what each rival expects before observing.
Post-hoc stories can make any result appear supportive.
Diagnostic check. What result would count against each rival?
4. Control other factors
Keep irrelevant conditions equal or account for them.
An uncontrolled cause can mimic the predicted difference.
Diagnostic check. What else could produce the outcome?
5. Measure the right outcome
Choose an observation linked directly to the disagreement.
A convenient measurement may not answer the contested question.
Diagnostic check. Does the outcome actually separate the claims?
6. Compare with predictions
Match the observed result to the pre-stated alternatives.
Interpretation should follow the test logic rather than preference.
Diagnostic check. Which prediction survived?
7. Handle inconclusive results
Allow both, neither or ambiguous outcomes.
Evidence need not force a choice when the test lacks resolution.
Diagnostic check. What remains undecided?
8. Bound the conclusion
Limit rejection or support to the domain and conditions tested.
One successful discrimination does not prove every broader claim.
Diagnostic check. How far can the result safely extend?
Worked examples: make the decision visible
Pause before each answer. State the controlling evidence in one line, then complete the operation. The explanations deliberately contrast a valid route with a plausible but invalid one, because transfer grows when the learner can say why a tempting route does not apply.
Worked example 1: Character motive
Task. A character leaves after an argument. Rival motives are anger and fear.
Reasoning. Leaving fits both; a detail about avoiding danger or expressing resentment would discriminate.
Answer. The departure alone is inconclusive.
Independent check. Find evidence tied to motive.
Contrast-and-transfer pass. Keep the governing principle from “Character motive”, but change one surface detail in the task: A character leaves after an argument. Rival motives are anger and fear. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—The departure alone is inconclusive.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 2: Number rule
Task. Rules A and B both output 6 for inputs 1,2,3 but differ at input 4.
Reasoning. Testing another agreement input adds no discrimination.
Answer. Use input 4 and compare predicted outputs.
Independent check. Write both predictions first.
Contrast-and-transfer pass. Keep the governing principle from “Number rule”, but change one surface detail in the task: Rules A and B both output 6 for inputs 1,2,3 but differ at input 4. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Use input 4 and compare predicted outputs.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 3: Plant lamp
Task. A lamp is moved closer and bubbles increase.
Reasoning. Both light intensity and temperature may change.
Answer. Control temperature before attributing the difference to light.
Independent check. Measure the competing factor.
Contrast-and-transfer pass. Keep the governing principle from “Plant lamp”, but change one surface detail in the task: A lamp is moved closer and bubbles increase. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Control temperature before attributing the difference to light.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 4: Material test
Task. Two explanations predict that a sample floats in water.
Reasoning. The float result supports neither uniquely.
Answer. Choose a liquid or measurement where predicted behaviour differs.
Independent check. The new test must remain relevant and safe.
Contrast-and-transfer pass. Keep the governing principle from “Material test”, but change one surface detail in the task: Two explanations predict that a sample floats in water. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Choose a liquid or measurement where predicted behaviour differs.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 5: Grammar choice
Task. Two connectors both fit grammar but encode cause and contrast differently.
Reasoning. A sentence with only one relation can separate them.
Answer. Paraphrase the clause relationship before choosing.
Independent check. Meaning, not appearance, is measured.
Contrast-and-transfer pass. Keep the governing principle from “Grammar choice”, but change one surface detail in the task: Two connectors both fit grammar but encode cause and contrast differently. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Paraphrase the clause relationship before choosing.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 6: Fair but weak
Task. Identical setups repeat an observation both rivals predict.
Reasoning. The method is fair and repeatable but non-discriminating.
Answer. Redesign around the rivals’ disagreement.
Independent check. Fairness and decisiveness are separate qualities.
Contrast-and-transfer pass. Keep the governing principle from “Fair but weak”, but change one surface detail in the task: Identical setups repeat an observation both rivals predict. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Redesign around the rivals’ disagreement.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 7: Unexpected result
Task. The apparatus gives neither predicted outcome.
Reasoning. The alternatives may be incomplete or the test may have failed.
Answer. Check method and consider another explanation.
Independent check. Do not force a winner.
Contrast-and-transfer pass. Keep the governing principle from “Unexpected result”, but change one surface detail in the task: The apparatus gives neither predicted outcome. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Check method and consider another explanation.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 8: Null difference
Task. No measurable difference appears where one rival predicts a large effect.
Reasoning. If sensitivity and controls are adequate, that rival loses support.
Answer. Report the tested limit cautiously.
Independent check. Confirm the instrument could detect the expected change.
Contrast-and-transfer pass. Keep the governing principle from “Null difference”, but change one surface detail in the task: No measurable difference appears where one rival predicts a large effect. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Report the tested limit cautiously.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 9: Scope limit
Task. One material behaves as predicted under room conditions.
Reasoning. The result settles that tested case only.
Answer. Do not claim all materials behave identically.
Independent check. State sample and conditions.
Contrast-and-transfer pass. Keep the governing principle from “Scope limit”, but change one surface detail in the task: One material behaves as predicted under room conditions. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—Do not claim all materials behave identically.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
Worked example 10: Reverse design
Task. Given two explanations, invent the best observation.
Reasoning. Choose the largest safe predicted difference with few confounds.
Answer. The test targets disagreement directly.
Independent check. Pre-register the possible interpretations.
Contrast-and-transfer pass. Keep the governing principle from “Reverse design”, but change one surface detail in the task: Given two explanations, invent the best observation. Before recalculating or rewriting, name which evidence remains controlling and which detail has changed. The original conclusion—The test targets disagreement directly.—may remain valid only if the changed detail does not alter the trigger, relationship or scope. Use the independent check again rather than trusting visual similarity. This second pass trains selection: the learner must decide whether the same method still applies, adapt one step when a condition changes, and refuse the method when its trigger disappears.
The method lab: change one condition at a time
For pairs of claims, mark agreement cases and disagreement cases. Choose the cheapest safe disagreement case as the test.
Write prediction tables with rows for outcomes and columns for rival explanations. Identify which outcome eliminates which rival.
Redesign a fair but non-discriminating test so the measured result bears differently on the alternatives.
Add a third explanation after an unexpected result and state a new discriminating prediction.
After a delay, design tests in English, Mathematics and Science without being told which evidence is decisive.
Common misconceptions and targeted repairs
Misconception 1: More evidence is always more decisive
What the learner may think. The learner counts observations.
Why it fails. Repeated agreement evidence may not separate rivals.
Targeted repair. Target disagreement.
Misconception 2: Fair means decisive
What the learner may think. The learner equates control with discrimination.
Why it fails. A fair test may still yield the same prediction from both claims.
Targeted repair. Check prediction contrast.
Misconception 3: Any supporting result proves one claim
What the learner may think. The learner ignores rival compatibility.
Why it fails. The same result can fit several explanations.
Targeted repair. Ask what the rival predicted.
Misconception 4: Predictions can be written afterwards
What the learner may think. The learner adjusts stories to the outcome.
Why it fails. Post-hoc flexibility weakens the test.
Targeted repair. Record predictions first.
Misconception 5: One winner proves a universal theory
What the learner may think. The learner widens scope.
Why it fails. The result is bounded by sample and conditions.
Targeted repair. State the tested domain.
Misconception 6: Neither prediction means the test is useless
What the learner may think. The learner forces a binary choice.
Why it fails. Unexpected results can expose missing alternatives or apparatus problems.
Targeted repair. Audit method and expand the set.
Graded independent practice with explained answers
Attempt each item without looking ahead. Write the evidence or relationship before the final response. Early items check recognition; middle items require explanation; later items change the surface form so that copying a worked example is insufficient.
Practice 1: Definition
Question. What makes a test decisive?
Explained answer. Rivals predict different observable outcomes.
Why this earns trust. Discrimination is explicit.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Rivals predict different observable outcomes.” from memory, but to recreate a justified answer from the new evidence.
Practice 2: Agreement
Question. If both predict the result, what follows?
Explained answer. Neither is uniquely supported.
Why this earns trust. The result is non-discriminating.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Neither is uniquely supported.” from memory, but to recreate a justified answer from the new evidence.
Practice 3: Timing
Question. When should predictions be written?
Explained answer. Before the test.
Why this earns trust. Post-hoc adjustment is limited.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Before the test.” from memory, but to recreate a justified answer from the new evidence.
Practice 4: Control
Question. Why control other factors?
Explained answer. They could mimic the predicted difference.
Why this earns trust. Causal interpretation is protected.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “They could mimic the predicted difference.” from memory, but to recreate a justified answer from the new evidence.
Practice 5: Mathematics
Question. Which input is useful?
Explained answer. One where candidate rules give different outputs.
Why this earns trust. Agreement inputs are weak.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “One where candidate rules give different outputs.” from memory, but to recreate a justified answer from the new evidence.
Practice 6: English
Question. What detail distinguishes motives?
Explained answer. Evidence linked differently to each motive.
Why this earns trust. Action alone may fit both.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Evidence linked differently to each motive.” from memory, but to recreate a justified answer from the new evidence.
Practice 7: Science
Question. What if light and heat both change?
Explained answer. Control or measure one before attributing the effect to the other.
Why this earns trust. Confounding is removed.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Control or measure one before attributing the effect to the other.” from memory, but to recreate a justified answer from the new evidence.
Practice 8: Null
Question. Can no difference be informative?
Explained answer. Yes, if the test was sensitive and one rival predicted a difference.
Why this earns trust. Method adequacy is checked.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Yes, if the test was sensitive and one rival predicted a difference.” from memory, but to recreate a justified answer from the new evidence.
Practice 9: Unexpected
Question. What if neither prediction occurs?
Explained answer. Check apparatus and consider missing explanations.
Why this earns trust. No winner is forced.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Check apparatus and consider missing explanations.” from memory, but to recreate a justified answer from the new evidence.
Practice 10: Scope
Question. How far may the conclusion extend?
Explained answer. Only to supported samples and conditions.
Why this earns trust. Overclaiming is blocked.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Only to supported samples and conditions.” from memory, but to recreate a justified answer from the new evidence.
Practice 11: Fairness
Question. Is a fair test automatically decisive?
Explained answer. No.
Why this earns trust. It must also separate predictions.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “No.” from memory, but to recreate a justified answer from the new evidence.
Practice 12: Routine
Question. State the sequence.
Explained answer. Rivals, disagreement, predictions, controls, observation, comparison and scope.
Why this earns trust. The logic is reproducible.
Transfer variation. Close the answer, wait several minutes, and rebuild the reasoning with different names, values or wording while preserving the same decision. Start by stating the evidence that activates the method; then produce the response and a separate check. If the new surface detail changes the governing condition, explain why the earlier answer pattern must not be copied. The aim is not to repeat “Rivals, disagreement, predictions, controls, observation, comparison and scope.” from memory, but to recreate a justified answer from the new evidence.
A four-session mastery cycle
Session 1: construct. Read the mechanism once, cover it and write a three-line model containing the trigger, the operation and the check. Add one example and one near-miss. Compare only after the model is complete. If the operation is right but the trigger is missing, the knowledge is unsafe because it may be applied in the wrong place.
Session 2: discriminate. Mix six items: two that clearly require this method, two that tempt the method but violate a condition, and two that require another skill. For every item, write “use” or “do not use” before answering. Accuracy in method selection matters more than speed at this stage.
Session 3: explain. Give a complete answer to an imagined classmate. Include the smallest sufficient evidence, the mechanism and the independent check. Remove any sentence that merely repeats the question or adds an unsupported claim. A concise causal explanation is stronger than a longer collection of familiar words.
Session 4: perform. Attempt a timed mixed set without notes. Mark uncertain decisions, but continue. During review, classify each error as trigger, evidence, operation, scope, unit or check. Reattempt one changed version after a delay. Mastery is shown when the learner chooses and adapts the method without needing the original example beside it.
Delayed transfer: prove that the learning lasts
Transfer test 1: Mystery rule
Separate two functions that agree initially.
Success criterion. A divergent input is chosen.
Transfer test 2: Narrative motives
Test rival character interpretations.
Success criterion. Evidence bears differently on the motives.
Transfer test 3: Experimental design
Remove a confound and preserve prediction contrast.
Success criterion. Fairness and discrimination both hold.
Transfer test 4: Delayed design
Create a new test one week later.
Success criterion. Predictions are stated before outcomes.
Examination control
1. Name the target. State the exact meaning, quantity, mechanism or decision requested.
2. Expose the condition. Write the grammatical, numerical or evidential condition that activates the method.
3. Perform visibly. Show the smallest sufficient reasoning step rather than a slogan.
4. Check the nearest trap. Test the tempting alternative that differs by one condition.
5. Present within scope. Keep units, domain, certainty and evidence boundaries in the final answer.
Under time pressure, compress the method to name rivals → find disagreement → predict → control → observe → compare → bound. The compressed routine is useful only after the full reasoning is understood. If it produces uncertainty, expand the step instead of guessing.
A compact error log
Use labels such as rivals vague, agreement case tested, predictions written late, confound uncontrolled, wrong outcome measured, ambiguity forced or scope widened.
Reattempt by choosing a new disagreement case rather than collecting more of the same evidence.
Notes for parents and tutors
Ask “what would each explanation predict?” before revealing an outcome.
Celebrate an honest inconclusive answer when the test cannot discriminate.
Use safe everyday comparisons and number rules.
Praise learners who narrow claims to the tested conditions.
Frequently asked questions
What is decisive evidence?
Evidence that bears differently on rival claims.
Can both predict the same result?
Yes, making that result inconclusive between them.
Why predict first?
To prevent changing the story after observing.
Is a fair test enough?
No.
What is a confound?
Another changing factor that can produce the outcome.
Can neither rival win?
Yes.
Does one test prove everything?
No.
Fastest check?
Ask where the rivals predict different results.
Next routes
Return to the complete volume directory to locate the series, use the original PSLE Learning Guide for the foundation route, and continue through PSLE English, PSLE Mathematics, PSLE Science for verified subject guidance. The next useful task is a mixed set in which this idea is sometimes relevant and sometimes irrelevant; that is where the learner must choose a method rather than imitate one.