Wait, what? A bulb does not light. Is the bulb faulty, is the circuit incomplete, is a connection unsuitable, or is the energy source not functioning as expected? A plant grows poorly. Is one input insufficient, is transport affected, is the observation period too short, or is the comparison itself unfair?

Many Primary 6 Science questions become difficult not because the student knows no explanation, but because several explanations initially seem possible.

The mature scientific task is not to pick the first familiar explanation. It is to compare the candidates against the evidence and choose the account that survives best.

This preserved Hougang Science Tutor P6 URL now owns that specific job: competing explanations. The old duplicated 2019–2020 tuition advertisement, obsolete schedules, location conflicts, A*/A1 promises and irrelevant image stack have been removed. The page is now a public Science reasoning satellite.

It is intentionally distinct from the other Hougang Primary 6 pages already rebuilt. Those cover PSLE triage, structured answers, experimental evaluation, multi-evidence integration, corrections, retrieval and checking under uncertainty. This page begins with a narrower problem: when two or more explanations fit at first glance, how do we decide which one deserves the conclusion?

Do not confuse “possible” with “supported”

A scientifically possible explanation is one that could happen in principle. A supported explanation is one that fits the actual evidence in the question.

For example, a bulb can fail because the bulb is faulty. But if the question explicitly states that the same bulb lights in another complete circuit, that possibility is weakened or ruled out.

The difference is crucial:

• Possible: the explanation can happen.
• Consistent: the explanation does not contradict the evidence.
• Supported: specific evidence points toward the explanation.
• Best supported: it explains the whole evidence set better than the alternatives.

Primary 6 reasoning improves sharply when students stop treating these states as identical.

Generate a small candidate set

When a question is genuinely ambiguous, generate two or three plausible explanations rather than a long list of everything that could ever happen.

A useful candidate set should:

• fit the topic and system;
• be compatible with at least some of the evidence;
• make different predictions somewhere;
• be testable against information in the question.

The purpose is not brainstorming for its own sake. It is to prevent premature commitment to the first explanation that comes to mind.

The discriminating-evidence principle

The best evidence is often not the most dramatic fact. It is the fact that different explanations predict differently.

Suppose Explanation A predicts the bulb should light when moved to a known complete circuit, while Explanation B predicts it still should not. Testing that condition separates the explanations directly.

The key question is:

What observation would one explanation expect but the other would not?

This turns evidence gathering into a search for discrimination rather than accumulation.

Use an explanation matrix

Evidence	Explanation A	Explanation B
Observation 1	Fits?	Fits?
Observation 2	Fits?	Fits?
Changed condition	Explains?	Explains?
Prediction	What should happen next?	What should happen next?
Contradiction	Any evidence against?	Any evidence against?

This is a teaching scaffold. In an examination, the learner may perform the same reasoning mentally.

The matrix prevents one attractive clue from dominating the whole question.

A good explanation must account for the awkward evidence too

Students naturally notice evidence that supports their first idea. The stronger test is the uncomfortable fact.

Ask:

• What evidence is hardest for my explanation to account for?
• Can the model explain it without inventing unsupported assumptions?
• Does another explanation handle it more naturally?

The best explanation should not require the learner to ignore inconvenient parts of the question.

Contradiction can eliminate an explanation quickly

In many PSLE Science questions, it is faster to rule out an explanation than to prove another one completely.

Look for contradictions:

• The explanation predicts an increase, but the repeated data shows a decrease.
• The explanation requires an open pathway, but the diagram shows a closed path.
• The explanation assumes a different starting condition, but the question states they were the same.
• The explanation needs a factor to change, but the method kept it constant.
• The explanation depends on a property the material is stated not to have.

One genuine contradiction can be more decisive than several weak supporting clues.

Explanation versus description

Two candidates may not actually be competing explanations. One may simply restate the observation.

For example:

• “The plant grew less because its height increased less.” — description repeated as explanation.
• “The plant grew less because the changed condition reduced a required process, which reduced the resources available for growth.” — mechanism.

Before comparing explanations, make sure each candidate contains a causal or functional account rather than a restatement.

The best explanation should connect the cause to the observation

A useful explanation has three major properties:

• Fit: it matches the evidence.
• Mechanism: it explains how the changed condition produces the observed outcome.
• Reach: it accounts for the important observations without needing a separate story for every one.

If one explanation fits one data point but fails the rest, while another explains the whole pattern through one coherent mechanism, the second deserves more confidence.

Prefer fewer unsupported assumptions

Students sometimes rescue a weak explanation by inventing extra facts:

• “Maybe the temperature changed.”
• “Maybe the instrument was faulty.”
• “Maybe the plant was already unhealthy.”

These are possible in real life, but unless the question provides evidence, they should not be used casually to save an explanation.

A stronger rule is:

Use the explanation that fits the evidence with the fewest unsupported additions.

This is not a demand for simplistic answers. It is a discipline against overfitting.

Competing explanations in experimental questions

Suppose two setups produce different outcomes. The first explanation is that the intended changed variable caused the difference. The second is that another uncontrolled condition caused it.

To choose between them:

1. State what the experiment was meant to test.
2. Check whether the intended variable changed as planned.
3. Check whether other important factors were controlled.
4. Inspect whether the measured outcome matches the question.
5. Use repeated or comparative evidence where available.
6. Decide whether the causal interpretation is strong or confounded.

Experimental design is therefore part of explanation selection.

Competing explanations in graph questions

A graph may show two quantities changing together. Several explanations can fit that association.

Ask:

• Did the experiment deliberately change one of the quantities?
• Were alternative factors controlled?
• Does the scientific model explain the direction of the relationship?
• Could a third factor affect both quantities?

The graph provides pattern evidence. The method determines how strongly that pattern can support a causal explanation.

Competing explanations in circuit faults

Circuit questions are ideal for diagnostic reasoning because several faults can produce the same visible outcome.

If a bulb does not light, possible candidates include:

• incomplete path;
• faulty bulb;
• unsuitable or missing connection;
• energy source problem.

The learner should use each extra observation to eliminate possibilities.

For instance, if the bulb lights in another known working circuit, the faulty-bulb explanation becomes much weaker. If replacing the battery changes nothing but closing a gap makes the bulb light, the evidence points elsewhere.

Science diagnosis works by narrowing possibilities.

Competing explanations in biological systems

A biological symptom can have several possible causes. The learner should stay within the syllabus and use the evidence given rather than importing medical or advanced biological assumptions.

Useful questions include:

• Which required input changed?
• Which transport or exchange process changed?
• Which structure is affected?
• Which whole-system outcome follows?
• What observation would distinguish input shortage from pathway failure?

The evidence should choose among the syllabus-level mechanisms.

Competing explanations in ecosystem questions

A population change in a food web can arise through several connected routes.

If one population decreases, possible explanations may involve:

• less food availability;
• more predation;
• competition;
• environmental change;
• another interaction specified by the question.

The correct explanation must be grounded in the given food-web relationships and conditions. Do not turn one generic ecosystem fact into the answer without tracing the actual connections shown.

Prediction is a powerful explanation test

A good explanation should make a useful prediction.

For each candidate:

• If this explanation is correct, what else should I observe?
• If I change one condition, how should the outcome respond?
• Which result would make the explanation hard to maintain?

When two explanations predict different outcomes, the next observation becomes decisive.

Negative evidence can rule out an explanation

If Explanation A requires a particular effect and that effect is absent under a method capable of detecting it, confidence in A should decrease.

Examples:

• a pathway is claimed to be open, but no downstream transport occurs;
• a changed factor is claimed to increase an outcome, but repeated values remain unchanged;
• a material is claimed to have a property, but the specified test repeatedly shows otherwise.

Absence is informative only when the expected effect should have been observable.

Do not overfit one strange result

One unusual observation can tempt the learner to invent a new explanation immediately.

Before doing so:

• check whether the result repeats;
• check the measurement method;
• check whether a condition changed;
• compare the result with the broader pattern.

A new explanation should account for more evidence than the old one, not just one anomalous point.

When the evidence cannot choose yet

Sometimes two explanations remain equally consistent with the information provided.

The scientifically mature response is not to pretend certainty.

State:

• which explanations remain possible;
• why the current evidence does not distinguish them;
• what additional observation or comparison would help.

In an examination, the question may still be designed so one answer is supported by all the given evidence. But the learning habit should remain: certainty must be earned.

MCQ distractors are often competing models

In multiple-choice questions, wrong options are often not random. They may represent:

• a common misconception;
• a reversed causal direction;
• a correct fact applied to the wrong condition;
• a conclusion drawn from the wrong comparison;
• a partial explanation missing one key link.

Instead of asking only “Which option is right?”, ask:

• What model would make each option seem right?
• Which evidence contradicts that model?
• Which option survives every condition?

This transforms elimination into model comparison.

Structured answers need the winning explanation, not the losing alternatives

Competing explanations are primarily an internal reasoning tool.

Once the evidence selects the best explanation, the final structured answer should usually present:

1. the relevant observation or condition;
2. the scientific mechanism;
3. the requested conclusion.

Do not write a long essay listing every discarded alternative unless the question explicitly asks for evaluation or comparison.

The internal reasoning can be rich while the final answer remains concise.

The best-explanation checklist

• Does it fit every important observation?
• Does it explain the changed condition?
• Does it contain a valid mechanism?
• Does it contradict any stated fact?
• Does it require unsupported assumptions?
• Does it make a prediction consistent with the data?
• Can a competing explanation fit the evidence equally well?
• What evidence would change the choice?

This checklist builds disciplined commitment.

Model limits: the simplest explanation is not always the correct one

Preferring fewer unsupported assumptions does not mean always choosing the shortest sentence.

A complex system may genuinely require several interacting causes. The objective is not simplicity at any cost. It is sufficient explanation without unnecessary invention.

If the evidence clearly shows two factors changed and both matter, a one-factor explanation may be too simple.

Good model selection balances fit, mechanism and economy.

Misconception checkpoint: “the first plausible explanation is probably right”

Ask the learner:

• What is your first explanation?
• What is one plausible alternative?
• Which evidence supports each?
• Which evidence contradicts each?
• What one observation would separate them most strongly?

The goal is not endless doubt. It is a brief disciplined check before commitment.

Five Primary 6 competing-explanation failure modes

1. First-answer lock

The learner commits to the first familiar explanation. Repair by generating one plausible alternative before finalising.

2. Possibility-equals-proof

The student argues “it could be X” as if that means X is supported. Repair by asking which evidence points specifically to X.

3. Favourite-evidence picker

Supporting evidence is noticed while contradictions are ignored. Repair with the awkward-evidence test.

4. Assumption rescuer

Extra facts are invented to save a weak explanation. Repair by separating given evidence from imagined possibilities.

5. Endless-alternative thinker

The learner keeps generating possibilities and never commits. Repair by using discriminating evidence and choosing the best-supported account once the evidence is sufficient.

A Phase 4 Primary 6 explanation-selection lesson

• Observe: state the evidence packet.
• Generate: form a small set of plausible explanations.
• Predict: state what each explanation expects.
• Compare: test each candidate against every relevant observation.
• Contradict: eliminate candidates that conflict with the evidence.
• Discriminate: find the most informative evidence.
• Assumption audit: remove explanations requiring unsupported additions.
• Select: choose the account with the best overall fit.
• Bound: keep confidence proportional to the evidence.
• Communicate: write the winning explanation concisely.

The learner becomes better at scientific judgment without becoming trapped in indecision.

Why small groups are useful for competing explanations

Three students may propose three different explanations for the same result. The tutor can turn that disagreement into a scientific comparison.

• What does each explanation predict?
• Which evidence supports it?
• Which evidence is hardest for it to explain?
• What new observation would separate the candidates?
• Which explanation requires the fewest unsupported assumptions?

The group learns to let evidence arbitrate between ideas.

What parents can practise at home

• When the child gives an explanation, ask for one plausible alternative.
• Ask what evidence would distinguish them.
• Ask which fact contradicts the weaker explanation.
• Ask whether any part of the answer was invented rather than given.
• Ask what the chosen explanation predicts next.
• Ask whether the same evidence could fit another model.
• Encourage commitment once one explanation clearly fits better.

The goal is disciplined comparison, not chronic uncertainty.

What evidence to bring when explanation selection is the bottleneck

• an MCQ where two options seemed plausible;
• a circuit-fault question;
• an experimental question with a confounding possibility;
• a graph question where association was mistaken for cause;
• a biological-system question with several possible causes;
• teacher corrections;
• the learner’s original explanation;
• one question where the first answer changed after new evidence was noticed.

These examples reveal whether the child is generating, comparing or eliminating explanations effectively.

How to tell whether explanation selection is improving

• The learner generates a plausible alternative when needed.
• Possibility is distinguished from evidence support.
• Contradictory evidence is actively checked.
• Discriminating evidence is identified faster.
• Unsupported rescue assumptions decrease.
• Graph associations are not automatically treated as causal.
• MCQ distractors are recognised as alternative models.
• Confidence becomes proportional to how well one explanation beats the others.
• The final structured answer remains concise after rich internal comparison.

These are signs that the learner is choosing explanations rather than merely recognising familiar phrases.

How this page fits the Hougang Science network

This eduKateSingapore page owns competing-explanation selection. It complements Integrating Multiple Pieces of Evidence, Scientific Checking, Confidence and Uncertainty, From Evidence to Complete PSLE Structured Answers, and Evaluating Evidence, Methods and Experimental Claims.

For the national subject map, continue to What Is Primary Science Education? | From Curiosity to Scientific Thinking, P3 to PSLE.

Official 2026 examination reference

For Standard Science examined in 2026, SEAB lists the revised PSLE Science subject as syllabus 0009. The official syllabus assesses Knowledge with Understanding together with Application of Knowledge and Scientific Inquiry, including interpretation, analysis, evaluation and communication of scientific reasoning. See PSLE Formats Examined in 2026 and the linked Science syllabus.

---

A strong Primary 6 scientist does not ask only, “Can this explanation be true?” The better question is, “Does it explain the whole evidence packet better than the alternatives, with a valid mechanism and without unsupported assumptions?”