Ages 8–12 · about 6 minutes
How Scientists Decide What Is True
How do I know if a big science story is real?
Do this first · Try this at home
Predict, count, check
Start with your hands. The reading makes far more sense afterwards.
You need: Paper and a pencil
0 of 3 steps done
What should I notice?
Notice: Your estimate may be near or far from the count. If the counts differ, check which dots each person included. This checks one drawing; it is not two independent physics experiments.
Now read on to find out why it happens.
The big idea
Write a prediction, check the measurement, then let others test it with new data.
- prediction
- A stated expectation you can compare with the result of a test.
- calibration
- Comparing a measuring instrument with a reference whose value is known with stated uncertainty.
- uncertainty
- How precisely a measurement is known, considering the limits of the instrument and method.
A prediction states what you expect before a test. Suppose you predict that twenty sheets of the same paper have about twice the mass of ten. Write that down, along with how you will measure. A prediction is something to check, not a result you already have.
Weigh each stack using the same tray and scale. Keep the paper type and conditions the same. Those controls help you compare the number of sheets without also changing the material. Record every reading, including one that surprises you.
Check the measurement
First check the empty tray and compare the scale with a reference mass near the masses you will measure. Calibration compares an instrument with a reference. A known-mass check can reveal some errors at that point; it cannot guarantee every reading is right.
Take several readings of each stack. Small differences reveal variation. A scale might also read consistently too high, so repeating alone cannot remove every error. Measurement uncertainty describes how precisely the result is known, taking account of the instrument, the reference and the method. It is more than the spread of repeated readings.
Make a fresh check
Another team can follow your written procedure with its own paper and equipment and collect new measurements. That is a replication attempt. Compare the results with their uncertainties, not just whether the last digit matches. A copied report adds no new measurement.
Separate teams can still share a faulty reference or the same software. Independence depends on how the measurements were made, not how many people repeat the story.
The explorer below models two equally precise measurements. Move the slider toward shared errors: combining the measurements then reduces uncertainty less. The numbers describe this simplified model, not the probability that a science claim is true. The slider does not inspect any real experiment.
When two results count as one
Raise shared uncertainty. The measurements remain two reports, but their combined error stops shrinking as if they were independent.
Use the result
If the stacks do not match your prediction, check the procedure, measurement limits and explanation. If they do match, that supports the prediction under these conditions; it does not prove every possible claim about paper. Keep the prediction and the results together so someone else can follow what changed.