Did you know that a student can write “I knew the warm water would work faster” after an experiment and accidentally erase the most interesting part of scientific learning? A prediction written after the result is no longer a fair record of what the student expected. An explanation that appears in the observation column can conceal what was actually seen.
Separating prediction, observation and explanation means making three different kinds of statements: a prediction says what you expect before a test; an observation records what you actually notice or measure; and an explanation interprets the results using relevant scientific ideas and evidence. Keeping them distinct helps learners reason carefully even when their first prediction turns out to be wrong.
Three voices in one small investigation
The prediction uses future-looking language: “If I change this condition, I think this result will follow.” It may be based on prior learning but remains something to test. The observation uses measured or directly noticed language: “The recorded time was 43 seconds.” The explanation uses reasoning: “One possible reason is …, and the comparison supports this only to the extent that other conditions were controlled.”
A learner needs all three because they answer different questions. What did I expect? What happened? Why might it have happened? A prediction is not a promise; an observation is not automatically a conclusion; and an explanation should not be invented merely to make the prediction look correct.
Worked example: sugar dissolving in water
Completely fictional teaching data: Suppose a class plans to compare how quickly 5 g of similar granulated sugar disappears when stirred in 100 mL of water at different starting temperatures. The cool-water trials start near 20°C and the warm-water trials near 40°C. The figures below are invented for learning and are not measurements from a real experiment. Warm water should be handled with appropriate adult supervision.
Prediction, written first: “With the same sugar, water volume and stirring method, sugar will become no longer visibly detectable sooner in warmer water.” This is testable: it identifies a changed condition (starting water temperature) and an outcome (time to no visible crystals). It does not declare in advance that the hypothesis is true.
Illustrative observations: In the cool-water trials, the invented recorded times are 82 and 78 seconds. In the warm-water trials, they are 43 and 41 seconds. The mean cool-water time is (82 + 78) ÷ 2 = 80 seconds; the mean warm-water time is (43 + 41) ÷ 2 = 42 seconds. The sample shows a shorter recorded time for the warm condition in this fictional setup. It does not demonstrate what every real trial would produce.
Possible explanation: Under comparable conditions, temperature can influence the rate at which a solid dissolves in a liquid. Warmer water generally has faster molecular motion, which may contribute to more rapid mixing and dissolution. But the observations alone cannot exclude differences in stirring, crystal size, starting temperature drift or the judgement of when the last crystal vanished. They also do not prove that every solid dissolves more quickly in every warmer liquid.
Here is the key distinction: “Sugar dissolved faster because the water was warm” is an interpretation, not a direct observation. “The second cup took 41 seconds” is an observation in the imaginary dataset. Both are useful when they are labelled honestly.
A simple three-column method, without a fancy worksheet
- Before: Write the question, the condition to change, the result to observe and one prediction. State why the prediction seems plausible without claiming it is established.
- During: Record the same type of measurement for every trial, with units, starting conditions and unusual events. Do not silently replace awkward results.
- After: Compare the actual data with the prediction. Work out only the calculations needed, showing what is being averaged or compared.
- Explain: Use the most relevant scientific idea, connect it to the pattern and identify important alternative explanations.
- Check: Ask whether someone else could distinguish the original prediction from the actual observations using the notes alone.
This is a planning–monitoring–checking habit, not a ritual to fill three boxes. A teacher can allow drawings or a simple table for observations when appropriate, while still expecting the learner to separate the evidence from their interpretation.
How a teacher can diagnose a confusing lab report
If the student writes “The water was energetic” in an observation field, ask what was actually measured. If the student reports only that an experiment “worked”, ask what changed and how it was recorded. If their prediction exactly matches the result in unusually polished wording, ask whether it was written before or after the test rather than assuming dishonesty.
Another diagnostic is to hide the explanation section. Can the learner still describe the setup and data precisely? Then hide the numbers and ask what scientific model they would use to interpret the pattern. These separate checks can reveal whether the difficulty is record-keeping, calculation, conceptual understanding or connecting an idea to evidence.
The mistakes that turn an investigation into a story
- Editing the prediction after seeing the outcome. It may make a report look successful but removes evidence of genuine testing.
- Writing a cause in the observation column. “It was faster because of heat” mixes the outcome with a proposed mechanism.
- Leaving out units or starting conditions. “82 versus 41” is not useful unless we know what was counted or timed.
- Changing several factors at once. Different sugar amounts, cup shapes and stirring rates complicate any temperature comparison.
- Ignoring unexpected results. An anomaly may reflect measurement error, uncontrolled conditions or an idea worth testing again.
- Claiming universal truth from a few trials. Repetition, measurement care and limits matter even when the pattern fits the prediction.
An alternative case: when the prediction is wrong
Imagine a student predicts that two paper towels of different brands will absorb equal water because they look equally thick. In a small trial, one visibly absorbs more before dripping. The student’s first prediction has not “failed” as an act of learning; it has been challenged by an observation. The honest next step is to ask how absorbency was measured, whether the pieces were the same size and whether the outcome repeats. The student can then revise the explanation and design a better comparison.
A wrong prediction is often more educational than an unexamined correct one. The learner gets to practise changing their mind for a reason. It is still important not to praise a poorly controlled test as decisive evidence just because it produced a surprise.
What students and parents can try at home
For a safe, simple activity with appropriate supervision, use familiar objects such as different paper towels and water rather than unfamiliar substances or heated equipment. Ask the child to say the prediction aloud before testing, then record what can be directly seen or measured. Keep the explanation for afterwards. Parents can support the distinction without supplying the ‘correct’ conclusion: “What did you expect? What did you see? Which part is your explanation?”
A tutor can move from modelling the three statements to giving a partially completed record and finally to a fresh investigation where the student sets up the notes independently. This progression echoes the diagnosis-first, explanation-first approach of eduKate’s Clementi Mathematics tutorial reference: inspect the exact missing thinking step, practise deliberately, then verify independent work.
Observe progress without treating neat notes as proof
Use a short checklist after several investigations: was the prediction recorded before results, were observations specific and dated or timed where relevant, were units preserved, did the explanation refer to the evidence, and did the learner state a reasonable limitation? A neat three-column table alone is not success. A learner who independently notices that one condition was not held constant is making meaningful progress.
To test transfer, present a new investigation with a surprising result but no labels. Ask the student to sort three statements into prediction, observation and explanation, then justify each placement. Next, ask them to write those statements from scratch for a new question. That second task checks whether the distinction is understood rather than memorised.
Limits: scientific work is rarely as tidy as three columns
Real investigations can be iterative: early observations can motivate new predictions, and explanations may be revised as new evidence arrives. The three-part distinction is a learning scaffold, not a claim that scientists must always work in a fixed order. Measurements can be uncertain, controls imperfect and explanations provisional. Good science makes those limits visible.
The broader science-education idea is supported by the National Science Teaching Association’s discussion of scientific practices, including observation, analysis, explanation and argument from evidence. Those practices are useful background reading, not a substitute for checking the requirements of a particular school’s task.
Frequently asked questions
What if my child’s prediction was wrong? Keep it visible. Compare it with observations, investigate why and revise the explanation; do not rewrite history to earn a neat-looking result.
Can an observation include a number? Yes. A measurement with units and method is often much clearer than “It changed a lot.” But a measured number can still be uncertain or affected by the method.
Must every explanation state a single cause? No. Some patterns have multiple plausible contributing factors. It is better to say what the data support and what remains unresolved than invent a confident cause.
Teach curiosity to keep its records straight
The question is not whether the learner guessed correctly. It is whether they can tell expectation from evidence and evidence from explanation, then use the differences to learn. Continue with the Core Aim guide to prediction testing or browse the eduKate Education library for connected learning skills. A prediction invites investigation; an observation gives it something real to answer to; an explanation earns its confidence.