- Foundation
- About 14 min
- 5 practice questions
Predictions, evidence and checking claims
Turn a vague claim into a testable prediction, plan a fair test with measurable evidence, and use scientific questions to check everyday and social-media claims.
If something looks wrong, please tell us.
By the end of this lesson you can
- Explain how a scientific prediction differs from a guess.
- Turn a vague claim into a testable question that asks for measurable evidence.
- Plan a fair test and use questions to judge a claim you meet online.
One of the greatest strengths of science is that it can predict. Using well-tested laws and models, we can say how far a football will travel when kicked, how much gas a reaction will give off, or how a person's breathing will change when they start running - often before anyone has done the experiment. This lesson shows what makes a prediction scientific and how to use the same thinking to check claims.
Prediction versus guess
A guess has nothing solid behind it. A scientific prediction is a reasoned expectation based on evidence, past patterns and established ideas. It is stated clearly enough that observations can show whether it was right. When a prediction matches observations, our confidence in the idea grows. When it does not, scientists re-check their assumptions, models and measurements. Either way, we learn something.
Making a claim testable
A vague statement such as 'my plant looks unhappy, so it needs more water' cannot easily be tested. A testable version names something measurable and a comparison: 'If I water this plant 100 mL a day instead of 50 mL a day for two weeks, it will grow taller than an identical plant given 50 mL.' Good questions ask about measurable data and past patterns, not just yes or no.
The prediction cycle
- Observe Notice a pattern or hear a claim.
- Predict State what should happen, in measurable terms.
- Test Collect evidence with a fair test.
- Compare Does the evidence match the prediction?
- Revise Keep, refine or reject the idea, then observe again.
Then back to step 1: Observe
Worked example 1: designing a fair test
Claim: 'Seeds sprout faster if you soak them in warm water first.' Plan a fair test.
Working: The one thing we change (the variable we test) is the soaking: half the seeds soaked in warm water, half not soaked. Everything else must be kept the same: the same kind of seed, the same number in each group (say 20), the same soil, pot size, amount of watering, light and room. What we measure: the number of seeds that have sprouted each day. Using 20 seeds per group, not just one, reduces the effect of a single unusual seed. If the soaked group sprouts clearly earlier, the claim is supported; if not, it is not. The group that is not soaked is the control: it gives us something to compare with.
Worked example 2: sharpening a prediction
Claim: Tanvi says, 'The school water tank will run dry before Friday because it is so hot.' What questions turn this into a testable prediction?
Working: Useful questions ask for measurements and past patterns: How many litres does the tank hold, and how full is it today? How many litres were used each day in recent hot weeks, according to the meter or the tank level? When is the tank next refilled? Divide the water left by the daily use to see how many days it will last. These questions replace 'it is so hot' with data that can be checked - and on Friday, the prediction can be compared with what actually happened.
Even careful weather forecasts are reliable for hours or a few days but less certain further ahead. Weather depends on many changing factors, and tiny differences in today's conditions can grow into big differences later.
Checking claims you meet online
Messages often claim that something is harmful or helpful without evidence. Take the claim that food becomes harmful during an eclipse. Ask simple scientific questions: What physically happens in an eclipse? (The Moon's shadow falls on part of the Earth.) Does food left in a shadow go bad? Does the temperature change enough to spoil food in an hour or two? No known physical, chemical or biological process supports the claim. Asking 'how would this work, and what evidence is there?' is a powerful habit.
| Question to ask | Why it helps |
|---|---|
| What exactly is being claimed? | A vague claim cannot be tested. |
| How could it work? | Real effects have a physical, chemical or biological cause. |
| What evidence is offered? | Stories and forwarded messages are not measurements. |
| Was there a fair comparison? | Without a control group, results can mislead. |
| Can others repeat the result? | Reliable findings show up again when tested by others. |
A scientific prediction is a reasoned, measurable expectation that observations can confirm or contradict. Change one variable at a time, keep everything else the same, and compare with a control.
Key takeaways: predictions in science come from evidence and established ideas, not guesswork. Make a claim testable by naming what will be measured and what it will be compared with. A fair test changes one variable and keeps the rest the same, with a control group. When predictions fail, scientists re-examine their assumptions. The same questions help you judge claims in daily life.
Check your understanding
-
1. Which of these is a testable scientific prediction?
- Tomorrow will be a lucky day
- Plants given 8 hours of light a day will grow taller in three weeks than identical plants given 2 hours
- Grey skies look sad
- This cricket team deserves to win
Show answer
Answer: B. It names what will be changed (hours of light), what will be measured (height) and a comparison over a set time. The others cannot be measured or checked.
-
2. In a fair test of whether fertiliser helps bean plants grow, which group is the control?
- Plants given fertiliser and extra water
- Plants kept in the dark
- Identical plants grown the same way but without fertiliser
- The tallest plants
Show answer
Answer: C. The control is treated exactly the same except for the one factor being tested, so it shows what happens without the fertiliser.
-
3. A scientist's prediction does not match careful measurements. What should she do first?
- Re-examine her assumptions, model and measurements
- Publish the prediction anyway
- Delete the measurements
- Make the prediction vaguer so it cannot be wrong
Show answer
Answer: A. A failed prediction is useful information. The scientific response is to check the assumptions, the model and the measurements, then refine the idea.
-
4. Why are weather forecasts usually less reliable a week ahead than a few hours ahead?
- Forecasters stop measuring after the first day
- Weather has no causes
- Instruments only work in the morning
- Tiny differences in present conditions can grow into large differences over time
Show answer
Answer: D. Weather depends on many changing factors. Small differences in today's conditions can grow over days, so long-range forecasts are less certain.
-
5. A message says a certain food becomes poisonous during an eclipse. Which question is most useful for checking it?
- How many people have forwarded the message?
- What physical, chemical or biological change during an eclipse could make food harmful?
- Does the sender seem confident?
- Is the message written in capital letters?
Show answer
Answer: B. Asking how the effect could happen points to evidence. An eclipse is a shadow; no known process makes food harmful in that time. Popularity or confidence is not evidence.
Sign in to keep track of the lessons you finish.