Science practice combines reading evidence with deciding what a design can establish. Begin with axes, units and variables. A pattern in a table supports a claim within the measured conditions; it does not automatically establish a universal law or isolate a cause.
A reliable method
Read the variables and units
Identify what changed deliberately and what was measured. Compare matched conditions when more than one factor changes. A percentage-point difference is a subtraction, not a relative percentage change.
Separate controls from repetitions
Controls limit alternative explanations. Repeating a condition helps measure variability. Repeating a flawed comparison many times does not remove its confounding variables.
Test competing models
Find a condition where the models predict different outcomes. A measurement both models predict cannot distinguish them. Choose a test whose result would favour one explanation over the other.
Bound predictions
Interpolation estimates within the measured range. Extrapolation extends outside it and needs an explicit assumption. A non-monotonic table does not support assuming the outcome always increases.
Worked examples
At equal light and temperature, plants given 0, 5, 10 and 15 mg of fertiliser average 8, 12, 15 and 13 cm. Does every added 5 mg increase height?
- Compare neighbouring doses: changes are +4, +3 and −2 cm.
- The last step decreases height.
- That measured counterexample contradicts an always-increasing claim; it does not identify the best dose outside the tested range.
Answer: No: height falls between 10 mg and 15 mg.
Model A predicts gas production needs light; Model B predicts it needs warmth but not light. Which matched test best distinguishes them?
- Keep the solution and warm temperature the same.
- Compare one trial in light with one in darkness.
- Model A predicts a difference; Model B does not. Changing warmth and light together would make the result harder to interpret.
Answer: Compare warm, otherwise identical trials in light and darkness.
Now use the method yourself.
Mark this guide as read, then use the method in topic practice before the checkpoint.