How to Decide Whether a PSLE Science Investigation Needs a Starting Measurement Before the Test Begins

Wait, What? A Final Measurement Cannot Tell You How Much Something Changed If You Never Knew Where It Started

Two plants are placed under different conditions for one week.

At the end, Plant P is 18 cm tall and Plant Q is 15 cm tall.

Which plant grew more?

You cannot know from those final heights alone.

If P started at 17 cm and Q started at 10 cm, then Q grew much more. If both started at 8 cm, then P grew more.

A starting measurement is needed when the scientific question depends on change from an initial state, not merely on the final state.

This looks simple, but it is one of the most useful investigation-design habits a Primary 5 or Primary 6 learner can build.

Quick Answer

Ask:

“Does the question ask what the system is like at the end, or how much it changed from the beginning?”

If the question is about change, a starting measurement is often needed.
If the question is only about the final state, a starting measurement may be unnecessary.
If different specimens may already start differently, a baseline can help separate pre-existing differences from effects that happen during the test.
If the starting condition is guaranteed identical by the setup and directly controlled, a separate starting measurement may add little.

Use this route:

READ THE SCIENTIFIC QUESTION → IDENTIFY THE OUTCOME → ASK WHETHER CHANGE FROM START MATTERS → CHECK WHETHER STARTING STATES MAY DIFFER → CHOOSE A STARTING MEASUREMENT IF NEEDED → APPLY THE TEST CONDITION → MEASURE THE FINAL OR LATER STATE → COMPARE CHANGE, NOT JUST FINAL VALUE → KEEP THE CONCLUSION WITHIN THE EVIDENCE.

The Exact PSLE Science Learning Job This Guide Owns

This guide owns one learner job: how a Primary 5 or Primary 6 learner decides whether a PSLE Science investigation needs a baseline or starting measurement before the changed condition begins, so pre-existing differences are not confused with change caused during the test.

It does not replace the guides on comparing set-ups with different starting values, endpoint versus time-series measurement, or fair-test variables. Those pages remain canonical for their own jobs.

This page owns the earlier design decision:

Do I need to measure the starting state before I begin?

Why This Matters in the 2026 PSLE Science Frame

For examination from 2026, Standard PSLE Science assesses the 2023 Primary Science syllabus. Official assessment objectives include interpreting and analysing information, evaluating observations and methods, making predictions and hypotheses, and communicating scientific explanations and reasoning.

Evaluating a method includes asking whether the evidence collected can actually answer the scientific question. A final measurement can be perfectly accurate yet still be insufficient if the question is about change and the starting state was never recorded.

Final State and Change Are Different Scientific Quantities

Question type	Evidence usually needed
Which setup has the greatest final height?	Final height
Which plant grew the most?	Starting height + final height
Which container has the most water remaining?	Final amount may be enough if starting amounts were controlled equal
Which container lost the most water?	Starting amount + final amount, unless identical starting amount is explicitly guaranteed
Which object reached the highest final temperature?	Final temperature
Which object increased in temperature the most?	Starting temperature + final temperature

The same experiment can support different conclusions depending on which measurements were taken.

Worked Example 1 — Plant Height

Three similar plants are assigned to three conditions. Their final heights are:

Plant	Final height / cm
P	18
Q	16
R	14

Can we conclude P grew the most?

Not unless we know the starting heights or have a justified reason to know they started equally.

If starting heights were 17, 10 and 9 cm, the growth amounts would be 1, 6 and 5 cm.

The largest final value is not automatically the largest change.

Worked Example 2 — Water Loss

Two dishes are left for the same time. At the end:

P contains 60 g of water.
Q contains 75 g of water.

If the question states that both dishes started with exactly 100 g, you can calculate:

P lost 40 g.
Q lost 25 g.

A separate measured baseline may be unnecessary because the starting amount is already known and controlled.

If the starting amounts are not given, the final values alone do not tell you which dish lost more.

Worked Example 3 — Temperature Change

Two objects end at 45°C and 42°C.

Object P may have increased by only 2°C if it began at 43°C. Object Q may have increased by 20°C if it began at 22°C.

If the scientific question asks which warmed more, starting temperature matters.

Worked Example 4 — Final Outcome Is the Actual Question

An investigation asks which of three insulating materials keeps a container’s water at the highest temperature after 20 minutes, with all containers beginning at the same stated temperature.

The starting state is already controlled and known. The decisive evidence may be the temperature after 20 minutes.

Do not add measurements merely because “more data is better”. Collect the evidence the question needs.

Worked Example 5 — Pre-Existing Differences Between Specimens

Two similar leaves are placed under different conditions. One leaf already has a larger mass before testing.

If you compare final mass only, you may confuse the starting difference with what happened during the test.

Recording starting mass lets you compare mass change.

Worked Example 6 — A Baseline Can Be an Observation, Not Only a Number

Before treatment, all leaves may be recorded as green with no visible wilting. After treatment, one group shows visible wilting.

The baseline is qualitative. It establishes that the observed state was not already present.

A starting measurement does not always mean measuring in centimetres or grams. It means recording the relevant starting state.

When a Starting Measurement Is Especially Valuable

Different specimens may naturally begin at different sizes or masses.
The question asks how much something increased or decreased.
The outcome is change in temperature, height, mass, volume, length or another quantity.
The test compares before and after states.
Natural variation is large enough that final values alone may mislead.
The starting condition cannot be guaranteed exactly by the method.

When a Separate Starting Measurement May Be Unnecessary

The starting quantity is directly fixed and stated.
The question asks only about final state.
The starting state is physically identical by construction and scientifically irrelevant to the outcome being compared.
The variable of interest is whether an event happens at all, not how much change occurred—provided other starting conditions are controlled appropriately.

“Measure everything before starting” is not a good universal rule.

Baseline Does Not Mean Control Set-Up

These are different ideas.

Baseline: the starting state before the test or intervention.
Control/reference set-up: a comparison condition used to interpret the effect of the changed factor.

An investigation can need both.

Baseline Does Not Mean Repeated Measurement Over Time

A baseline is one starting measurement.

A time series measures repeatedly during the process.

If the question is only about total change, start and end may be enough. If the question asks when change occurs, how rate changes or whether a turning point appears, repeated measurements may be needed.

The Baseline Decision Test

What exact quantity or state is the question asking about?
Is it a final state or a change?
Could the specimens/set-ups already differ before testing?
Is the starting state explicitly controlled or given?
Would two different starting states produce the same final value but different changes?
If yes, record the baseline.

Why Final Values Can Hide the Scientific Story

Imagine two runners finish at the same line. One started 10 metres away; the other started 100 metres away.

The final position alone cannot tell you who travelled farther.

Many Science measurements work the same way. Final value and amount of change are different properties.

Do Not Subtract Values From Different Objects Without Checking Identity

If the starting height belongs to Plant A and the final height belongs to Plant B, the difference is not the growth of one plant.

Before calculating change, preserve object identity:

Same object or comparable matched unit → starting measurement → later measurement → change.

Do Not Turn Baseline Differences Into Proof the Groups Were Unfair

Small natural differences can exist even when specimens are chosen carefully.

The scientific question is whether those differences matter for the comparison and whether the method accounts for them appropriately.

Do Not Assume Equal Final Values Mean Equal Change

If P and Q both finish at 20 units:

P may have started at 10;
Q may have started at 19.

Same final state. Very different change.

Do Not Assume Different Final Values Mean the Tested Condition Caused the Difference

If P ends higher than Q but also started higher, the final difference may be partly or entirely pre-existing.

Use change-from-baseline evidence before attributing the difference to the test condition.

The Earliest-Weak-Link Diagnostic

Failure signature	Earliest weak link	Repair
Chooses largest final value as largest change	Final state confused with change	Record/inspect starting value
Subtracts measurements from different specimens	Object identity lost	Match start and end to same case
Always asks for baseline even when start is fixed equal	Method improvement used mechanically	Ask what evidence the question needs
Uses baseline but ignores control condition	Baseline confused with comparison design	Separate starting state from reference set-up
Measures start and end but question asks about rate over time	Measurement schedule too sparse	Add intermediate time points if rate/pattern matters
Sees different final values and claims causation	Pre-existing difference not excluded	Compare changes and fair-test conditions

Misconception Repair — “The Highest Final Value Means the Biggest Increase”

No. Increase is final value minus starting value for the same relevant case.

Misconception Repair — “A Baseline Is Always Necessary”

No. If the starting state is already fixed, known and irrelevant to the scientific question, an extra measurement may not improve the evidence.

Misconception Repair — “More Measurements Automatically Make the Investigation Better”

Measurements must answer the scientific question. Unnecessary measurements can add workload without improving the conclusion.

Misconception Repair — “The Baseline Proves the Cause”

A baseline helps separate starting state from later change. Causal conclusions still depend on fair comparison, controlled conditions, relevant measurement and evidence quality.

How This Appears in Investigation-Planning Questions

Before proposing measurements, translate the investigation question into an evidence requirement.

If the question asks:

“How does Condition X affect the increase in Y?”

you need evidence of Y before and after, or another valid way to determine the increase.

If it asks:

“How does Condition X affect the final value of Y after 10 minutes?”

the final measurement may be the direct target, provided starting conditions are controlled appropriately.

How This Appears in Data Questions

A table may provide starting and final values. Do not ignore the starting column merely because the final values look easier to compare.

Ask what the question requests:

final amount;
total change;
percentage change;
rate;
difference between groups.

Use only the level of arithmetic appropriate to the question and learner stage.

How This Appears in Method-Evaluation Questions

A method may compare final outcomes of naturally varying specimens without recording their initial state.

A strong evaluation says why this matters:

Without the starting measurement, a final difference could have existed before the test, so the method cannot cleanly determine how much change occurred during the investigation.

The improvement should preserve the original scientific question: record the relevant starting state before applying the test condition.

The Baseline Practice Sequence

Classify ten questions as final-state or change questions.
For each change question, identify the needed starting state.
Use examples where starting values are equal and examples where they differ.
Use qualitative baselines as well as numerical ones.
Separate baseline from control set-up.
Separate baseline from repeated time-series measurement.
Use a flawed method and propose the smallest repair.
Return later with unfamiliar contexts.

Unfamiliar Transfer Challenge

Three rubber bands are tested under different conditions. Their final lengths are 15 cm, 17 cm and 19 cm.

Question: Which rubber band increased in length the most?

Can you answer?

No. You need the initial lengths for the same rubber bands or valid evidence that their starting lengths were identical.

Change the question to:

“Which rubber band had the greatest final length?”

Now the final data answer it directly.

The scientific job—not the topic—decides whether baseline evidence is needed.

Delayed Independent Return

Three to five days later, take a new investigation and answer:

What exact outcome is being investigated?
Is it final state or change?
Could the starting states differ?
Is the starting state already fixed or given?
What baseline would be relevant?
Would one baseline plus one final value be enough?
Would repeated measurements over time be needed instead?
What conclusion would the data support?

The Investigation-Checking Receipt

Did I identify whether the question asks about final state or change?
Did I preserve the same object/specimen across starting and final measurements?
Could pre-existing differences explain the final result?
Is the starting state explicitly controlled or known?
Did I avoid adding an unnecessary baseline mechanically?
Did I distinguish baseline from control set-up?
Did I distinguish baseline from a time series?
Does my method collect the evidence the scientific question actually needs?
Does my conclusion use change rather than final value when appropriate?

Evidence and Model Limits

In professional science, baseline design can become statistically complex. Researchers may use matched groups, repeated measures, random assignment, covariate adjustment and other methods. None of that advanced machinery is required for the learner job here.

The Primary-level principle is simpler:

If your claim depends on how much something changed, make sure you know where the relevant thing started.

A baseline alone does not prove causation and does not fix every method weakness. It simply protects the investigation from confusing starting state with later change.

Useful Internal Routes

Parent and Tutor Teaching Guide

Use pairs of questions that look nearly identical:

“Which plant is tallest at the end?”
“Which plant grew the most?”

Ask the learner why the second question needs information the first may not.

Then use the same numbers with different starting values. This helps the child see that the data requirement comes from the scientific question, not from a memorised rule that “investigations always need a before reading”.

When the child proposes a baseline, ask:

What exactly will you measure?
Why does the question need it?
How will it be matched to the later measurement?
What mistake would happen if you did not record it?

A good method improvement should have a scientific reason, not merely add apparatus or data.

Authoritative and Research References

These sources support broader scientific inquiry and evidence reasoning. They do not prescribe a PSLE-specific rule that every investigation must collect a baseline.

The Quiet Ending

Before you ask how much something changed, know where it began.

Before you blame a test condition for a final difference, check whether the difference was already there.

A good starting measurement is not extra decoration.

It is the zero point of the scientific story you are trying to tell.

Small Group Tutorials