PSLE Science Reality Lab Vol No.325 | “Control = 100%” — Does 150% Mean the Treatment Had 150 Units?

PSLE-SCI-REALITY-0325

Wait, What? A Graph Can Say 150% Even When Nobody Measured 150 of Anything

A scientific graph shows three bars. The control bar is labelled 100%. Treatment A is 150%. Treatment B is 80%. A student looks at the graph and says, “Easy. The control had 100 units, Treatment A had 150 units and Treatment B had 80 units.”

That interpretation can be completely wrong even if every number on the graph is correct.

Researchers often normalise data to a reference. They may take the control value, define it as 100%, and express every other result relative to that reference. If the raw control measurement was 40 arbitrary units and Treatment A was 60, the normalised graph can show 100% and 150%. If another experiment had a raw control of 200 units and a treatment of 300, its normalised graph can also show 100% and 150%.

The two graphs look identical after normalisation even though the absolute measurements are very different. That makes “Control = 100%” a perfect Reality Lab object. The learner must ask what the percentage is a percentage of, reconstruct the hidden baseline, and separate relative comparison from absolute amount.

Quick Answer

On a normalised graph, 100% can be a chosen reference value rather than an absolute measurement of 100 physical units.
150% of control usually means the reported value is 1.5 times the chosen control reference under the stated calculation.
It does not automatically mean the treatment measured 150 grams, 150 cells, 150 millimetres, 150 joules or 150 of any other physical unit.
To recover an absolute value, you need the original reference measurement and the exact normalisation rule.
Two studies can both report 150% of control while having very different raw measurements.
Normalisation can make relative comparisons clearer, but it can also hide the absolute scale, baseline variation and some details about the raw data.
The safe habit is: find the reference, find the equation, then decide what the percentage can support.

The Exact Learner Job This Page Owns

This page owns one real-world evidence-transfer job: evaluating a scientific graph that sets a control or reference to 100% without mistaking the normalised percentage for an absolute amount.

It does not own generic percentage calculation, graph reading or reference-line interpretation. Those remain with the existing PSLE Science skill owners. Reality Lab applies those skills to a common scientific communication move: turning raw measurements into values relative to a reference so different treatments can be compared more easily.

Original Reality Lab Case: Two Experiments, the Same 150%

This is an original constructed example. No published experiment, examination question or competitor graphic has been copied.

Two fictional laboratories study a light-sensitive pigment. Both laboratories make graphs in which their own control is normalised to 100%.

Experiment	Raw control	Raw Treatment A	Normalised control	Normalised Treatment A
Lab 1	40 fluorescence units	60 fluorescence units	100%	150%
Lab 2	200 fluorescence units	300 fluorescence units	100%	150%

After normalisation, the two comparisons look identical. In each experiment the treatment is 1.5 times its own control. But the raw measurements differ by a factor of five.

If a student sees only the normalised graph, the student can say that Treatment A is reported as 150% of that experiment’s control. The student cannot recover the raw fluorescence value unless the original control value or another absolute reference is given.

The Basic Calculation Behind “Control = 100%”

A common normalisation is:

normalised value = measured value ÷ reference value × 100%

If the control measures 40 units and Treatment A measures 60 units:

60 ÷ 40 × 100% = 150%.

Notice what disappeared from the displayed number. The raw unit—perhaps grams, counts per second, fluorescence units, centimetres or another measurement—is no longer visible. The graph now expresses a ratio relative to the reference.

This can be useful because it puts the comparison into a common relative form. But the reader must remember that the 100% baseline was created by a calculation. It was not necessarily measured as “100”.

Observed, Measured, Normalised, Claimed and Inferred

Layer	What happened
Observed	The instrument or method produced raw observations under stated conditions.
Measured	Those observations were converted into quantities such as mass, intensity, concentration, length or count.
Normalised	Each measured value was divided by a chosen reference and rescaled, often so the reference became 1 or 100%.
Claimed	The graph may show that a treatment is higher or lower than the reference by a relative amount.
Inferred	The reader may infer biological, physical or chemical meaning—but only as far as the experimental design and evidence support.

Normalisation sits between measurement and interpretation. It changes the representation, not the underlying experiment.

The Denominator Check: What Exactly Became 100%?

Every percentage needs a denominator. “150%” is incomplete until we know 150% of what.

150% of the untreated control?
150% of the starting value?
150% of a calibration standard?
150% of a long-term average?
150% of the maximum possible response?
150% of a value measured in the same specimen before treatment?

These can all produce percentages that look similar while answering different scientific questions. The label “Control = 100%” should therefore send the learner searching for the methods, caption or legend that defines the reference.

The Hidden-Baseline Check: Same Percentage, Different Reality

Imagine two plant experiments. Both report leaf growth as a percentage of their own control.

Study	Control growth	Treatment growth	Normalised result
A	2 cm	3 cm	150%
B	20 cm	30 cm	150%

Both studies show the same relative change. They do not show the same absolute growth. A reader comparing only the normalised bars could miss this difference completely.

That does not make normalisation dishonest. It means the representation has a job: emphasising a ratio. If the reader’s question is about absolute size, the raw values matter too.

The Comparison Check: Can We Compare 150% Across Different Studies?

Not automatically. Two papers can each report 150% of control while differing in almost everything else: species, temperature, concentration, instrument, duration, control condition, raw scale and uncertainty.

A cross-study comparison first asks whether the same quantity and a meaningfully comparable reference were used. If one study normalises to an untreated control and another normalises to a pre-treatment baseline, the percentages may answer different questions.

This is a powerful PSLE Science habit: matching-looking numbers are not automatically matching measurements.

The Unit Check: Where Did the Units Go?

Suppose a sensor originally measured a signal in volts. After dividing every value by the control voltage, the ratio may be dimensionless. Multiplying by 100 turns it into a percentage. The original physical unit has been cancelled in the ratio.

That cancellation is mathematically valid when the same kind of quantity is divided by itself. But it also means the graph no longer tells you the original voltage. A student who writes “150 volts” because the bar says 150% has reintroduced a unit that the graph does not provide.

The Zero and Near-Zero Problem

Ratios become difficult when the reference is zero or very close to zero. Dividing by zero is undefined. Dividing by a very small uncertain value can make a normalised percentage enormous and unstable.

Imagine a control measurement of 0.1 units and a treatment of 0.2 units. The treatment is 200% of control. That sounds dramatic, yet the absolute difference is only 0.1 unit. If the measurement uncertainty itself is around that size, the giant percentage may exaggerate how decisive the evidence feels.

This is why scientists inspect raw values, uncertainty and method limits instead of treating a percentage alone as the complete story.

The Variability Check: Did Every Control Equal the Same Value?

In a real experiment, control measurements may vary. Researchers might normalise each experimental run to its matched control, or they might divide all samples by one mean control value. These choices can be reasonable, but they are not identical.

A bar at 150% does not tell you, by itself, whether every treatment replicate was exactly 1.5 times every control replicate. The plotted bar may be a mean of normalised values, and the spread around that mean matters.

Look for error bars, individual data points, sample size and a methods description. Normalisation can simplify the centre of the story while leaving the variability hidden unless the figure shows it.

Normalisation Can Reveal a Pattern That Raw Numbers Hide

Normalisation is not merely a source of confusion. It can solve real comparison problems.

Imagine four instruments whose absolute signal strengths differ because of unavoidable calibration or specimen differences, but each instrument responds proportionally to the same treatment. Expressing each response relative to its own control can make the shared treatment pattern easier to see.

The important distinction is this: normalisation may improve our view of relative change while reducing our view of absolute level. A good scientific figure makes the chosen purpose clear and gives enough information to recover what matters.

What Evidence Would Strengthen a Normalised Comparison?

The caption states exactly what was set to 100%.
The equation or method used for normalisation is clear.
The same measured quantity and reference definition are used across compared groups.
Raw values, or enough information to recover their scale, are available when absolute interpretation matters.
Individual points or uncertainty show how variable the data were.
The control is stable enough that dividing by it does not create an unstable ratio.
The conclusion stays relative when only relative data are shown.

What Would Weaken the Claim?

The graph says “% control” but never defines the control.
A treatment at 150% is described as “150 units” with no raw unit given.
Two studies with different controls are compared as though their percentages share one absolute scale.
The control is near zero, making the ratio highly unstable, but the uncertainty is hidden.
The plot omits all information about raw values, sample size and variation while making a strong absolute claim.
Different groups were normalised using different rules, but the bars are presented as directly comparable.

Worked Case 1: 150% of 40

A control is 40 units and the treatment is shown as 150% of control. If the calculation is treatment ÷ control × 100%, the treatment corresponds to 60 units. The 150 on the graph is not an absolute measurement.

Worked Case 2: Two Identical Percentages, Different Raw Differences

Study A: control 10, treatment 15. Study B: control 1,000, treatment 1,500. Both show 150%. The relative increase is the same. The absolute difference is 5 in A and 500 in B. Which description matters depends on the scientific question.

Worked Case 3: 200% Sounds Huge

A control signal is 0.05 and a treatment signal is 0.10. The normalised treatment is 200%. A headline says “Treatment doubled the response.” That statement may correctly describe the ratio, but it tells us nothing about whether a change of 0.05 is large, reliable or practically important without more context.

Worked Case 4: The Missing Raw Baseline

A figure shows Control 100%, Treatment X 120% and Treatment Y 80%, but no raw data. Can a learner calculate the number of cells in each group? No. The graph supports relative comparisons with the reference, not an absolute cell count.

Worked Case 5: Different References

Graph 1 normalises every sample to an untreated control. Graph 2 normalises every sample to its own starting value before treatment. Both contain a 130% bar. Those bars do not automatically express the same scientific relationship because their denominators differ.

Tempting Reasoning That Fails

“100% means the control measured 100 units.” It may simply have been defined as the reference value.
“150% means 150 physical units.” The units may have cancelled during normalisation.
“Two 150% bars are the same result.” Their raw baselines and methods may differ.
“A larger percentage always means a larger absolute difference.” The baseline size matters.
“Normalised data are manipulated, so they are unscientific.” Normalisation can be a legitimate representation when the rule and purpose are clear.
“The graph hides raw values, so the result must be false.” Missing raw scale limits certain conclusions; it does not automatically invalidate the relative comparison.

Model and Measurement Limits

Normalisation is a mathematical transformation. It cannot improve a poor experiment by itself. If the control is badly measured, biased or unstable, every value divided by that control inherits the problem. If the treatment and control were not comparable, setting one to 100% does not repair the design.

Normalisation can also compress information. Once the raw scale is removed, two very different experiments can look alike. The reader therefore needs the methods and, where possible, the raw or absolute data to decide how far the graph can travel.

How Far Can the Conclusion Travel?

If a graph clearly defines the control as 100%, a bar at 150% can support the conclusion that the reported treatment value was 1.5 times the chosen reference under that normalisation rule, subject to the experiment’s uncertainty and design.

It does not automatically support an absolute value, a cross-study comparison, a claim that every replicate rose by 50%, or a claim that the treatment caused the change unless the experimental comparison justifies causation.

PSLE-Style Transfer Case

A fictional experiment measures enzyme activity. The graph states “Control = 100%”. Treatment P is shown at 160%. A pupil writes, “Treatment P had an enzyme activity of 160 units.”

Question: Explain why the statement is not justified.

Reasoned answer: The graph shows a value relative to a control that has been normalised to 100%. The 160% means Treatment P is reported as 1.6 times the control under the stated calculation. The graph does not provide the original control activity in physical units, so the absolute enzyme activity cannot be concluded from the percentage alone.

Explained Practice

Practice A: Control = 20 mg, Treatment = 30 mg. What is the treatment as a percentage of control? 150%. What is the treatment mass? 30 mg. The two answers describe different representations of the same measurement.

Practice B: A graph shows 80% of control. Does that mean the sample lost 20 physical units? No. It means the reported value is 20% below the reference on that relative scale. The absolute difference depends on the control value.

Practice C: Two experiments both show 120%. One uses a 5-unit control; the other a 500-unit control. Which has the larger absolute increase? The second, if the same kind of quantity and simple normalisation are used: increases are 1 and 100 units respectively.

Practice D: The methods say “each sample was divided by its own pre-treatment value”. What does 100% represent now? Each sample’s own baseline, not necessarily one common control group.

Practice E: The control is 0.01 ± 0.02 units and the treatment is 0.02 ± 0.02. The normalised value is numerically 200%. Should the percentage alone be treated as strong evidence of a large effect? No. The reference is very small relative to its uncertainty, so the ratio can be unstable and the raw uncertainty matters greatly.

Delayed Independent Return: R-E-L-A-T-E

R — Reference: What value was made 100%?
E — Equation: How was the normalised percentage calculated?
L — Like with like: Are the same quantity, method and conditions being compared?
A — Absolute scale: Is the raw measurement visible or hidden?
T — Treatment claim: Does the figure show relative change, absolute amount or both?
E — Evidence boundary: What cannot be concluded from the normalised graph alone?

Return to the graph a day later and use R-E-L-A-T-E without help. Then transfer it to a completely different object—a battery-capacity chart, a plant-growth graph or a fluorescence experiment—to test whether the reasoning habit survives a change of context.

Parent and Tutor Teaching Guide

Use three cups of water. Label Cup A “control” and pour 40 mL into it. Put 60 mL into Cup B and 32 mL into Cup C. Ask the learner to express B and C as percentages of A. The answers are 150% and 80%. Then hide the measuring cylinder and show only the percentages. Ask whether another set of cups containing 200, 300 and 160 mL could produce the same graph. The learner should discover that normalisation preserves ratios but can hide absolute scale.

Next, deliberately change the reference. Make Cup B the 100% reference. Ask how the other percentages change even though no physical amount changed. This is an important insight: the reported relative number can change because the reference changes.

Finally, ask the learner to write two sentences about a normalised graph: one sentence the evidence supports and one sentence it does not. For example: “Treatment A was 150% of the control” is supported; “Treatment A measured 150 units” is not supported without the raw control value. This keeps the lesson focused on evidence language rather than calculation alone.

Authoritative Sources

The Quiet Return

A bar at 150% can be completely correct without anyone measuring 150 of anything.

The number is telling you a relationship.

Before reading a percentage as an amount, find the value that was chosen to become 100%.

Small Group Tutorials