How to Turn a PSLE Science Results Table Into an Honest Graph Without Distorting the Data

Wait, What? A Correct Table Can Become a Misleading Graph

A learner completes an investigation and records the results correctly. The table is sound. The units are correct. The measurements are in the right rows.

Then the learner draws a graph.

The vertical scale jumps from 0 to 5 to 20 to 25. The plotted points are joined even though the tested conditions are unrelated categories. A point that was never measured is quietly added “to make the line smoother”. The graph now tells a different story from the table.

A graph is not decoration added after an investigation. It is a representation of evidence. If the graph changes the meaning of the table, the Science has changed on the page even though the experiment itself has not.

This guide teaches a Primary 5 or Primary 6 learner how to move from results table to graph while preserving the scientific meaning of every value, condition, unit and evidence limit.

Quick Answer

Before drawing, make six decisions.

What was deliberately changed? This identifies the condition or variable that organises the results.
What was measured or observed? This identifies the outcome to represent.
Are the conditions numerical and ordered, or are they categories? This affects which graph forms make scientific sense.
What are the units? Keep them attached to the quantities.
What scale shows the data clearly without changing the numerical relationships?
Which values were actually observed? Plot those faithfully and do not invent intermediate evidence.

Then check the return path:

RESULTS TABLE → VARIABLE ROLES → REPRESENTATION CHOICE → AXIS LABELS + UNITS → REGULAR SCALE → PLOT OBSERVED VALUES → CONNECT ONLY WHEN THE RELATIONSHIP JUSTIFIES IT → READ THE GRAPH BACK → CHECK EVERY POINT AGAINST THE TABLE.

The Exact PSLE Science Learning Job This Guide Owns

This guide owns one learner job: how a Primary 5 or Primary 6 learner turns PSLE Science results into an appropriate graph while preserving the changed condition, measured outcome, labels, units, scale, observed values and limits of the evidence.

It does not replace the existing guides on reading graphs, understanding measurement resolution, distinguishing trends from single comparisons or interpreting data gaps. Those pages teach how to read or reason from representations. This page owns the construction step: how to make a graph that remains faithful to the evidence that produced it.

Why This Matters in the Current 2026 PSLE Science Frame

For examination from 2026, Standard PSLE Science assesses the 2023 Primary Science syllabus. The official assessment objectives include knowledge with understanding, application of scientific facts, concepts and principles, and scientific inquiry. Inquiry includes interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.

Graphs are therefore not separate from scientific thinking. They are one way evidence can be organised and communicated. A learner who constructs a graph must preserve what the investigation actually measured and what the data can support.

This guide is a learning framework, not an official SEAB graph-marking rubric. If a question gives specific graph instructions, follow the instructions in that question.

First Principle: The Table Owns the Evidence

Your graph does not create new measurements. It rearranges measurements that already exist.

Suppose a results table contains:

Time / min	Temperature / °C
0	70
5	62
10	56
15	52

The graph may make the cooling pattern easier to see, but it must not quietly add a value at 7 minutes, change 56°C to 55°C because that makes the line smoother, or make the 5-minute interval look twice as wide as the next 5-minute interval.

Every plotted point should be able to travel back to a specific table entry or stated observation.

Step 1 — Find the Changed Condition and Measured Outcome

Before thinking about axes, identify the scientific roles.

Example:

Exposed wet surface area / cm²	Mass of water lost in 20 min / g
20	2
40	4
60	7
80	9

The deliberately varied condition is exposed wet surface area. The measured outcome is mass of water lost after the same time.

This matters because a graph that swaps those roles may still contain the same numbers but answer a different reading question. The learner should know what relationship the graph is meant to show before drawing it.

Step 2 — Decide Whether the Conditions Are Ordered Numbers or Categories

Not all first columns behave the same way.

Numerical or ordered condition

Examples include time, temperature, length, number of layers or amount of water. These conditions have an order, and the numerical spacing can matter.

Categorical condition

Examples include material P, material Q, wood, metal, plastic, or three species. The labels name different groups; they are not points on one numerical scale.

This distinction prevents a common error: drawing an apparently continuous rising line through categories merely because the printed table lists them in a certain order.

Choosing a Representation: Match the Graph to the Scientific Structure

If the question specifies the representation, follow it. If you are practising how to choose, use the structure of the variables rather than a memorised rule such as “Science always uses line graphs”.

Data structure	Representation that may be useful	Scientific caution
Categories compared by one measured value	Separate bars or a clear table	Do not imply meaningful numerical spacing between category names.
Measured outcome across ordered numerical conditions	Points on numerical axes; a line may help show the relationship when appropriate	Do not invent measurements between observed points.
Outcome measured repeatedly over time	Time-series points, often connected to show sequence	A connecting line shows a path through the measured sequence, not proof of every exact intermediate value.
Very few values where exact numbers matter	Table may remain clearest	Do not force a graph when it adds no useful interpretation.

The word may matters. Good representation depends on the question and data, not one universal school rule.

Step 3 — Put Variables and Units Where They Can Be Read

A label should name the quantity, and the unit should tell the reader how the quantity is measured.

Weak label: Water

Better label: Mass of water remaining / g

Weak label: Time

Better label: Time / min

A learner who writes the quantity and unit explicitly is less likely to confuse total amount, amount changed and rate later.

Step 4 — Build a Regular Numerical Scale

On a numerical axis, equal physical spacing should represent equal numerical change unless a special scale is explicitly used and clearly marked.

Consider this invalid scale:

0 — 5 — 10 — 30 — 35

If those labels are placed at equal distances, the 10-to-30 jump has been visually compressed while the 5-unit jumps are expanded. The representation distorts the numerical relationship.

A suitable regular scale might use 0, 5, 10, 15, 20, 25, 30, 35 at equal spacings, or another consistent interval that fits the data clearly.

Does an Axis Have to Start at Zero?

Not every scientific graph must begin at zero. A restricted range can sometimes make small changes easier to see. But if you use a non-zero start, it must be clearly labelled, and you must remember that the visual difference can look larger than the numerical difference.

The relevant learner job is honesty of scale: do not let the picture make a small difference look like a huge scientific effect.

Worked Example 1 — Time and Temperature

Original data:

Time / min	Temperature / °C
0	80
5	71
10	64
15	59
20	55

A faithful graph would preserve the equal 5-minute spacing and place each temperature at the correct y-value.

What should not happen?

putting 0, 5, 10, 20, 15 in time order;
making 0–5 occupy the same page width as 5–15;
plotting 60 instead of 59 because it is neater;
adding an unmeasured 25-minute point;
calling the curve “proof” of the exact temperature at every second.

The graph shows the observed cooling pattern. It does not give measurements that were never taken.

Worked Example 2 — Material Categories

Suppose three materials are tested for the amount of water absorbed:

Material	Water absorbed / mL
P	8
Q	3
R	6

P, Q and R are categories. The alphabet does not mean the material changes continuously from P through Q to R. Separate bars can compare their measured outcomes without implying a continuous path between categories.

If you rearranged the categories R, P, Q, the scientific values would still be 6, 8 and 3 mL. The order of category names does not create a scientific trend.

Worked Example 3 — Unequal Tested Intervals

Data are collected at 0, 2, 5 and 10 minutes.

A common graphing error is to place those four time points equally far apart because there are four rows in the table.

That changes the time geometry. The gap from 0 to 2 minutes should be smaller than the gap from 5 to 10 minutes on a regular numerical time axis.

Equal row spacing in a table does not mean equal numerical spacing on a graph.

Worked Example 4 — Data With a Missing Measurement

Suppose measurements exist at 0, 5 and 15 minutes, but no reading was recorded at 10 minutes.

Do not add a 10-minute point by averaging neighbouring values unless the question explicitly asks you to estimate it. The graph should preserve the gap in observed evidence.

You may sometimes connect measured points to show the overall time sequence, but the connecting line is not a hidden measurement. Keep that distinction clear.

Worked Example 5 — A Qualitative Outcome

An investigation records a colour as pale, medium or dark rather than measuring a numerical intensity.

Do not automatically assign 1, 2 and 3 and then calculate a numerical slope. The categories may have an order, but the distance between pale and medium is not necessarily the same as the distance between medium and dark.

A clear table or category-based representation can preserve the actual evidence without inventing precision.

Step 5 — Plot Only What Was Observed or Calculated Legitimately

A plotted point should come from one of two places:

a direct observation or measurement supplied by the data;
a derived value the question legitimately asks you to calculate from those measurements.

A calculated change is not the same as a direct measurement. If you graph a derived quantity, label it as that derived quantity.

Step 6 — Decide Whether Connecting Points Adds Meaning or Invents It

Joining points can be helpful when the x-axis represents an ordered process such as time or a numerical test condition and the line is used to help the reader follow the measured sequence.

But joining points does not prove:

that the change between them was perfectly straight;
that no turning point occurred between sparse measurements;
that the process continued beyond the tested range;
that categories between P and Q somehow exist.

Ask: Does the line help communicate the ordered relationship, or does it suggest measurements or continuity the investigation did not establish?

Step 7 — Read the Graph Back Into Words

After drawing, ignore the table for a moment and read the graph as a new reader would.

What is on the horizontal axis?
What is on the vertical axis?
What do the units mean?
Which condition gives the highest recorded outcome?
Where does the measured quantity increase, decrease or remain similar?
Are there gaps or unmeasured intervals?

If the words you read from the graph do not match the table, something went wrong in construction.

The Table-to-Graph Return Check

Graph feature	Return question
Axis label	Can I point to the corresponding table heading?
Unit	Is it the same unit used by the evidence?
Scale	Does equal spacing represent equal numerical change?
Point	Which exact table entry produced this point?
Bar	Which category and measured value does it represent?
Connecting line	Am I using it to follow an ordered relation rather than invent exact hidden data?
Trend statement	Does it stay within the measured range?

Failure Signature 1 — The Graph Has No Scientific Quantity Labels

Axes say “x” and “y”, or simply “results”.

Earliest weak link: variables were not identified before drawing.

Repair: write changed condition + unit and measured outcome + unit before choosing a scale.

Failure Signature 2 — Equal Page Spacing Represents Unequal Numbers

The learner labels equal tick gaps 0, 10, 20, 50, 60.

Earliest weak link: scale meaning is not stable.

Repair: choose one interval and repeat it regularly.

Failure Signature 3 — The Graph Invents a Missing Point

A neat point appears at a condition that was never measured.

Earliest weak link: observed data and estimated data were mixed.

Repair: mark only observed values unless an estimate is explicitly requested, and label estimates as estimates.

Failure Signature 4 — Categories Are Treated as a Continuous Number Line

P, Q and R are connected as though Q is numerically halfway between P and R.

Earliest weak link: category structure was not identified.

Repair: represent categories as categories and compare their measured outcomes directly.

Failure Signature 5 — The Scale Makes a Tiny Difference Look Huge

Values 98, 99 and 100 occupy nearly the full height of a graph, and the learner concludes one condition is “many times greater”.

Earliest weak link: visual height replaced numerical comparison.

Repair: compare the actual numbers and state the size of the difference.

Failure Signature 6 — The Learner Chooses a Graph Before Reading the Variables

“I always use a line graph in Science.”

Earliest weak link: representation choice became a memorised ritual.

Repair: identify whether the changed condition is numerical/ordered or categorical and what relationship the graph needs to communicate.

Misconception Repair — “A Graph Is More Scientific Than a Table”

No. A table can be the clearest way to preserve exact values. A graph is useful when it makes a comparison, pattern or relationship easier to see.

The scientific quality lies in faithful evidence and appropriate interpretation, not in choosing the most visually impressive representation.

Misconception Repair — “Smooth Means Accurate”

A smooth curve can be visually attractive and scientifically unjustified. Sparse measurements do not prove a smooth path between them.

Accuracy means faithfulness to evidence and method, not aesthetic neatness.

Misconception Repair — “Every Point Must Be Connected”

No universal rule says every plotted point in every primary Science graph must be connected. Whether connection is meaningful depends on the variable structure and the question.

If the question specifies how to graph, follow it. Otherwise ask what the line would mean scientifically.

Misconception Repair — “The Biggest Graph Feature Is the Biggest Scientific Effect”

Visual size depends on scale. Scientific magnitude comes from the numbers and units.

The Earliest-Weak-Link Diagnostic

Visible error	Earliest weak link	Repair
Axes swapped without understanding	Variable roles unclear	State changed condition and measured outcome first
Missing units	Quantity meaning incomplete	Copy quantity + unit from evidence before plotting
Irregular numerical intervals	Scale construction weak	Choose one regular interval
Incorrect point	Table-to-point transfer error	Check coordinate pair against one row at a time
Invented intermediate point	Observation/estimate boundary lost	Plot measured evidence only
Line through unrelated categories	Category/numerical distinction lost	Reclassify the x-variable
Correct graph but wrong trend statement	Representation-to-reasoning return failed	Read actual values and condition range back in words

A Construction Protocol for Practice

Read the scientific question.
Circle the changed condition.
Underline the measured outcome.
Write units beside both.
Decide whether the condition is numerical/ordered or categorical.
Choose the representation that fits the relationship or follow the question’s specified format.
Choose regular numerical scale intervals.
Plot one row at a time.
Check each point or bar against the table immediately.
Add only scientifically meaningful connections.
Read the completed representation back into words.
Check that no conclusion exceeds the observed range.

Practice Sequence — From Table Fidelity to Independent Representation

Copy fidelity: plot five simple ordered points from a table.
Unit fidelity: use a table where the units change between examples.
Interval fidelity: use unequal x-values such as 0, 2, 5 and 10.
Category discrimination: compare material categories and refuse to turn alphabet order into a numerical trend.
Missing-data discipline: include one unmeasured condition and keep it unplotted.
Scale challenge: choose a scale for values clustered narrowly without exaggerating the interpretation.
Representation choice: decide whether a table, separate bars or an ordered numerical plot communicates the evidence best.
Transfer: repeat in an unfamiliar Science context with no graph-type hint.

Unfamiliar Transfer Challenge

A fictional material changes stiffness when placed at four temperatures: 10°C, 20°C, 40°C and 70°C. The measured stiffness values are 3, 5, 6 and 6 units.

Before drawing:

changed condition = temperature / °C;
measured outcome = stiffness / stated units;
temperature is numerical and ordered;
the x-gaps are 10, 20 and 30°C, so they should not be drawn as equal numerical intervals;
only four stiffness values were observed;
the equal values at 40°C and 70°C do not prove the true response was perfectly constant at every untested temperature between them.

If the graph preserves all six statements, the construction job is transferring.

Delayed Independent Return

Several days later, take a fresh results table and construct a representation without notes. Then answer:

What scientific relationship does my graph show?
Did I preserve every unit?
Are numerical spacings regular?
Can every plotted value be traced back to evidence?
Did I treat categories as categories?
Did I imply continuity that was not observed?
Would a reader reach the same numerical conclusions from my graph as from the table?

If the answers remain yes on a changed context, the skill is becoming independent.

The Graph-Construction Receipt

I know what was changed.
I know what was measured.
I know whether the conditions are numerical/ordered or categorical.
My quantities have units where relevant.
My numerical scale is regular and readable.
Every plotted point or bar matches the table.
I have not invented missing measurements.
I connect points only when the connection helps represent an ordered relationship.
I do not read visual steepness or height without checking the numbers.
My graph supports the same conclusions as the original evidence.

Useful Internal Routes

Parent and Tutor Teaching Guide

Begin with an intentionally simple table. Ask the learner to explain what each column means before drawing anything. If the child cannot name the changed condition and measured outcome, do not move to graph paper yet.

Next, create one deliberate trap at a time. Use unequal time intervals. Use categorical materials. Leave one measurement missing. Give values that sit near the top of a narrow range. Ask, “What would this graph accidentally imply if we drew it carelessly?”

When checking, avoid praising neatness first. Ask for the evidence return path: “Show me where this point came from in the table.” If the learner can do that for every point, the graph is becoming scientifically accountable.

Finally, ask the learner to read the graph back without seeing the table. If the reconstructed values and relationships match the original evidence, representation and Science are staying connected.

Authoritative and Research References

The research references support broader graphing and representation principles. They do not create a universal PSLE-specific graph-construction marking formula.

The Quiet Ending

The table is where the evidence arrived.

The graph is where the evidence becomes visible in another form.

Your job is not to make the pattern prettier than the experiment.

Your job is to let the evidence survive the journey.

Small Group Tutorials