Wait, what? A Primary 6 Science question can contain a diagram that suggests one answer, a table that seems to suggest another, and one sentence in the stem that changes the meaning of both.

Many students solve each piece separately and then choose whichever evidence they noticed first. Stronger scientific reasoning asks a different question:

What conclusion is supported when all the relevant evidence is considered together?

This preserved Hougang Primary 6 Science URL now owns that specific job: multi-evidence integration before committing to a conclusion. The old 2019–2020 duplicated tuition advertisement, location conflicts, A*/A1 claims and irrelevant image stack have been removed.

This role is intentionally distinct from the other Hougang Primary 6 Science pages. Those already cover PSLE triage, structured-answer construction, experimental evaluation, correction retention and checking under uncertainty. This page sits one step earlier than the final answer: how does the learner combine several different evidence sources into one coherent scientific judgment?

A Science question can contain several evidence channels

Primary 6 questions may provide information through:

• written conditions;
• labels in a diagram;
• relative positions;
• arrows or pathways;
• tables of values;
• graphs;
• before-and-after observations;
• experimental method details;
• known scientific concepts;
• results from more than one setup.

The student must decide which sources are relevant, whether they agree, and how they connect.

The final conclusion should be the output of that integration—not the first familiar fact that appears.

Build an evidence packet before building the answer

Before writing, collect the minimum evidence packet.

1. Task: What must be decided, predicted, compared or explained?
2. Condition: What changed or differs between the cases?
3. Observation: What happened?
4. Data: Which value, trend or comparison matters?
5. Model: Which scientific concept explains the relationship?
6. Boundary: What can the evidence support—and what can it not support?

Not every question needs all six explicitly written. The learner should mentally assemble enough of the packet to avoid premature commitment.

Do not let the diagram overpower the text

Diagrams are visually powerful. Students can assume that what looks larger, closer, brighter or thicker is scientifically meaningful even when the question never defines those features as data.

When text and diagram appear to conflict, ask:

• Is the diagram drawn to scale?
• Which features are explicitly labelled?
• Does the text define a condition that the drawing only represents schematically?
• Which source carries the more precise information?

A diagram is evidence only through the features it was designed to encode.

Do not let one dramatic data point overpower the trend

A single unusual result can attract attention, especially in a graph or table.

The learner should ask:

• What is the overall pattern?
• Does the unusual point repeat?
• Could it reflect measurement variation?
• Does it change the conclusion or merely reduce confidence?
• Would the answer be different if that point were rechecked?

Scientific judgment weighs the whole evidence set rather than cherry-picking the most memorable value.

Converging evidence: different routes point to the same conclusion

Evidence converges when independent pieces support the same explanation.

For example:

• a diagram shows a blocked pathway;
• a table shows reduced output downstream;
• the known system model predicts that blocking the pathway should reduce that output.

Each piece strengthens the same causal story.

The learner should be able to say:

The diagram establishes the changed condition, the data establishes the observed effect, and the scientific model connects the two.

This is evidence integration.

Conflicting evidence: do not choose your favourite

Sometimes two pieces of evidence appear inconsistent.

A table may show little difference even though the model predicts a strong effect. A diagram may imply a complete path while the observation shows no output. A repeated result may disagree with one isolated reading.

Do not choose the evidence that matches expectation. Diagnose the conflict.

• Are the sources measuring the same thing?
• Do they refer to the same time point?
• Was one condition different?
• Could one measurement be anomalous?
• Is the model too simple?
• Is the apparent conflict only a misunderstanding of scale or units?

The conflict itself becomes information.

The discriminating evidence: which fact separates two explanations?

When two explanations seem possible, not every piece of information is equally useful.

The strongest evidence may be the one fact that one explanation predicts and the other does not.

Ask:

• What would Explanation A predict?
• What would Explanation B predict?
• Which observation differs between those predictions?
• Do we have that observation in the question?

This is a powerful way to escape keyword matching.

Evidence can have different reliability

Two pieces of evidence should not always be weighted equally.

A repeated measurement under a controlled method may deserve more confidence than one rough estimate. A clearly labelled numerical value may outweigh visual size in a non-scale diagram. Several consistent trials may outweigh one unusual reading.

A Primary 6 learner can ask:

• Which evidence came from the most controlled method?
• Which was repeated?
• Which is most directly related to the claim?
• Which is vulnerable to a known measurement issue?

Integration is partly about weighting evidence, not merely collecting it.

Evidence relevance: true facts can still be irrelevant

A question may mention several scientifically true facts. Only some matter to the requested conclusion.

Ask:

• If I remove this fact, does my conclusion change?
• Does this fact distinguish the competing explanations?
• Does it establish a condition, an observation or a mechanism?
• Am I including it simply because it belongs to the chapter?

Evidence earns its place by contributing to the reasoning route.

Integrate before explaining

A common error is to explain each evidence source separately and never connect them.

For example:

• Sentence 1 describes the graph.
• Sentence 2 describes the diagram.
• Sentence 3 states a textbook fact.

All three may be correct, yet the answer still lacks synthesis.

A stronger structure is:

Because the diagram shows ______ and the data shows ______, the relevant scientific process is ______, which explains why ______.

The words are not a fixed template. The logic is: sources combine into one model.

Use a source-role map

Evidence source	Its role
Diagram	Shows structure, connection, direction or changed condition
Table/graph	Shows measured pattern, magnitude or comparison
Method	Tells us how trustworthy or causal the comparison may be
Observation	Shows what actually happened
Scientific model	Explains why the evidence should be connected
Question wording	Defines what conclusion is required

The learner does not need to draw this table in an examination. It is a teaching scaffold for understanding how different information sources contribute different jobs.

Temporal evidence: make sure the pieces refer to the same stage

A diagram may show the setup at the beginning while the graph shows measurements taken later.

Students can create false contradictions by comparing different time points as if they were simultaneous.

• What is the starting state?
• What happens during the interval?
• At what time was each measurement taken?
• Does the final diagram represent the same stage as the table?

Time is part of the evidence packet.

Conditional evidence: one sentence can change the whole interpretation

Words such as “except”, “only”, “after”, “before”, “same”, “different”, “most likely”, “kept constant” and “not drawn to scale” can alter the role of the visual or numerical evidence.

Teach the learner to mark the conditions that constrain interpretation.

A correct graph reading under the wrong condition is still the wrong answer.

Negative evidence: absence can matter

Sometimes what does not happen is informative.

• A bulb does not light despite one connection appearing correct.
• A measured outcome does not change when one factor is varied.
• An expected product or observation is absent.
• A pathway is present but no downstream effect occurs.

Absence can rule out or weaken an explanation—but only if the method would have detected the expected effect.

This is why method quality and evidence interpretation remain connected.

Do not count evidence pieces as votes

Three weak pieces of evidence do not automatically beat one strong discriminating observation.

Evidence is not a simple vote count.

The learner should evaluate:

• relevance;
• quality;
• independence;
• how directly the evidence distinguishes the explanation;
• whether several pieces are actually repetitions of the same underlying fact.

This is an age-appropriate beginning to evidence weighting.

The cherry-picking trap

Students can latch onto the one value or observation that supports their first answer and ignore contradictory information.

Use the opposition check:

• What evidence supports my answer?
• What evidence might challenge it?
• Can my explanation account for both?
• If not, which part needs revision?

A strong explanation survives contact with the inconvenient evidence too.

The consistency check across representations

The same scientific relationship may be represented as:

• a diagram;
• a table;
• a graph;
• a verbal description;
• a causal chain.

If the learner’s interpretation is correct, it should survive translation.

Ask:

• Can I turn the graph pattern into a sentence?
• Can I represent the sentence as a causal arrow?
• Does the arrow agree with the diagram?
• Does the predicted result match the table?

Representation consistency is a powerful error detector.

The minimum sufficient evidence test

Some students overload their working memory by trying to hold every detail in the question equally.

Ask:

What is the smallest set of evidence that is sufficient to justify the conclusion?

This does not mean ignore other information. It means identify the core discriminating evidence first, then check the remaining information for contradiction.

This reduces cognitive load and makes answer construction more efficient.

The contradiction-first strategy

When several answers look plausible, search first for the condition that makes one impossible.

In MCQ work, this can be faster than proving one option completely from scratch.

• Which option contradicts the diagram?
• Which contradicts the data direction?
• Which violates a stated condition?
• Which requires an unsupported assumption?
• Which cannot fit the known scientific model?

Evidence integration can work by elimination as well as construction.

From evidence packet to structured answer

Once the evidence is integrated, answer construction becomes simpler.

1. State the relevant observation or comparison.
2. Connect it to the scientific concept.
3. Explain the mechanism.
4. State the conclusion requested.
5. Keep the claim within the evidence boundary.

For deeper work on this final conversion step, use Hougang Primary 6 Science | From Evidence to Complete PSLE Structured Answers.

Model limits: evidence can support more than one reasonable hypothesis

Not every question provides enough information to choose one explanation with complete certainty.

When two explanations remain possible, the scientifically mature response is to identify what additional evidence would distinguish them.

• What new measurement would help?
• What comparison is missing?
• What control would rule out the alternative?
• What observation would one explanation predict but the other would not?

“Not enough evidence yet” can be a rigorous conclusion.

Misconception checkpoint: “the first clue tells me the answer”

Give the learner a multi-source question and ask:

• What does the diagram tell you?
• What does the table tell you?
• What does the method tell you about confidence?
• Which scientific concept connects them?
• Is there any evidence that challenges your first interpretation?
• What one fact would most strongly discriminate between two answers?

If the learner can answer these, they are integrating rather than reacting.

Five Primary 6 evidence-integration failure modes

1. First-clue committer

The student decides after noticing one familiar feature. Repair by assembling the minimum evidence packet before committing.

2. Diagram-dominant reader

The visual representation overrides explicit textual conditions. Repair by identifying which diagram features actually encode data.

3. Data-point cherry-picker

One convenient value is selected while the overall pattern is ignored. Repair with trend and opposition checks.

4. Equal-weight integrator

All information is treated as equally reliable and relevant. Repair by weighting evidence according to method quality and discriminating power.

5. Evidence-list writer

The student mentions diagram, table and concept but never connects them. Repair by building one integrated causal story.

A Phase 4 Primary 6 evidence-integration lesson

• Task: define the decision the question requires.
• Collect: identify relevant evidence from every representation.
• Role: decide what job each evidence source performs.
• Weight: judge relevance and reliability.
• Converge: identify evidence pointing to the same model.
• Challenge: actively search for conflicting evidence.
• Discriminate: find the fact that separates competing explanations.
• Integrate: build one coherent scientific mechanism.
• Bound: state only what the whole evidence packet supports.
• Communicate: convert the integrated model into a concise answer.

The learner moves from evidence collection to evidence synthesis.

Why small groups help with evidence synthesis

Three students may prioritise three different pieces of evidence. The tutor can ask each to defend the weighting.

• Why is that evidence relevant?
• What does it establish?
• What evidence challenges your interpretation?
• Which observation would change your conclusion?
• Can your model account for all the evidence, not just your favourite part?

The group learns that synthesis is a contest between explanations, not between personalities.

What parents can practise at home

• Ask the child to identify evidence from each representation before answering.
• Ask which piece is most discriminating and why.
• Ask what evidence challenges the first answer.
• Ask whether the diagram is to scale or merely schematic.
• Ask whether one unusual data point changes the whole pattern.
• Ask what extra evidence would distinguish two explanations.
• Ask the child to explain how the diagram, data and concept fit into one story.

The goal is to slow commitment just enough for the whole evidence packet to be seen.

What evidence to bring when synthesis is the bottleneck

• a question combining diagram and data;
• a question combining method and result;
• an MCQ with two plausible options;
• a structured question where the child mentioned evidence but did not connect it;
• teacher corrections;
• one question where the learner ignored a contradictory clue;
• the student’s original working or annotations;
• one example where a second piece of evidence changed the first conclusion.

These samples reveal whether the failure is extraction, weighting or integration.

How to tell whether multi-evidence reasoning is improving

• The learner delays commitment until relevant sources are inspected.
• Diagram features are distinguished from decorative presentation.
• Overall trends are considered before isolated values.
• Evidence relevance and reliability are weighed more explicitly.
• Conflicting evidence triggers diagnosis rather than cherry-picking.
• The strongest discriminating evidence is identified more quickly.
• Representations are translated consistently.
• Answers integrate evidence instead of listing it.
• Claims become better bounded when evidence remains ambiguous.
• The learner can state what additional evidence would resolve uncertainty.

These are signs that the learner is building scientific judgment rather than clue matching.

How this page fits the Hougang Science network

This eduKateSingapore page owns multi-source evidence integration. It complements Scientific Checking, Confidence and Uncertainty Under PSLE Conditions, 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

SEAB’s official PSLE Formats Examined in 2026 lists Science as a revised subject under code 0009. Students should always use the current official syllabus and school guidance for examination requirements rather than relying on an older tuition page.

---

A strong Primary 6 Science conclusion is rarely the product of one clue. Read the condition, inspect the diagram, compare the data, judge the method, apply the model, search for contradiction and commit only when the evidence packet tells one coherent story—or state clearly what evidence is still missing.