Coaching Centre
Variables and fair tests
Independent, dependent and controlled variables, how to design a fair test, why scientists repeat trials and average them, and what valid and reliable mean.
Justin wanted to know whether his cats preferred full-cream or low-fat milk. He gave one cat a small bowl of full-cream and the other cat a saucer of low-fat, then left the room. An hour later the low-fat milk was gone, so he decided his cats prefer low-fat. Can you see at least four reasons his conclusion means nothing? By the end of this page you will.
Three kinds of variable
A variable is anything that can change in an experiment and could affect the result. In a fair test you sort every variable into one of three jobs.
| Variable | What it is | Ask yourself |
|---|---|---|
| Independent variable (IV) | The one thing you change on purpose | "What am I changing?" |
| Dependent variable (DV) | What you measure to see the effect | "What am I measuring?" |
| Controlled variables | Everything else that could affect the result, kept the same | "What must I keep the same?" |
The independent variable is the cause; the dependent variable is the effect. Its value depends on what you did to the independent variable, which is an easy way to remember which is which. In the hypothesis "If salt is added to water, then the boiling temperature will rise", the amount of salt is the IV and the boiling temperature is the DV.
What makes a test fair
A fair test changes only the independent variable and keeps every controlled variable the same. If you change two things at once, you can't tell which one caused the difference.
Mia tests whether the type of surface affects how far a car rolls. Trial 1: carpet, ramp 20 cm high. Trial 2: tiles, ramp 30 cm high. The car rolls further on the tiles. Can she say tiles are the reason?
- Choose what you want to investigate, then set up Test A and Test B. The machine checks whether your comparison is fair before it tells you anything.
- Run the same set-up three times. Do you get exactly the same distance? Why not?
- Find out which variable has the biggest effect on distance.
Repeat it, then average
Even a fair test gives slightly different results each time: you release the car a little differently, the stopwatch is pressed a fraction late, the floor isn't perfectly even. That's why scientists do repeat trials, usually at least three, and calculate the average (mean).
A car rolled 152 cm, 148 cm and 156 cm in three trials from the same height. Find the average.
If one trial is far from the others (say 152, 148 and 95 cm), it is an outlier. Something probably went wrong in that trial, such as the car hitting a chair leg. Check it, repeat that trial if you can, and leave a clear mistake out of the average, saying so in your report.
Valid and reliable
Two words come up in every investigation question from now to Year 12:
- Valid: the experiment really tests what it says it tests. A test is valid when it is fair (only the IV changes and the controlled variables are kept the same) and measures the right thing.
- Reliable: repeating the experiment gives results that are close to each other. You check reliability by doing repeat trials, and by seeing whether other people get the same results.
Back to Justin. Different cats (one may just be hungrier), different amounts of milk, a bowl versus a saucer, no repeats, and nobody watching to see who drank what. The test isn't fair, so it isn't valid, and with one trial nobody knows if it's reliable either.
Test yourself
Students test whether the amount of sunlight affects how tall bean plants grow. The dependent variable is
Height is what they measure. Sunlight is what they change.
In the same experiment, which is a controlled variable?
Water could affect growth, so it must be kept the same for every plant.
Why should only one variable be changed in a fair test?
Change two things and you can't tell which one caused the effect.
Three trials gave 12 s, 14 s and 13 s. The average is
(12 + 14 + 13) ÷ 3 = 39 ÷ 3 = 13 s.
Repeating an experiment several times and getting similar results shows that the results are
Consistent results across repeats = reliable.
Trials of 54 cm, 51 cm, 18 cm and 53 cm. Which statement is best?
18 cm is far from the rest, so something likely went wrong in that trial.
Experiment: does the temperature of water change how long a sugar cube takes to dissolve? Sort each variable.
In a fair test only variable is changed. This is the variable. The variable that is measured is the variable. All other variables are . Doing at least three repeat and calculating the makes the results more .
Read Justin's milk experiment at the top of this page. Identify two reasons it was not a fair test, and describe how you would improve the method.
It was not fair because each cat got a different milk, so one cat may just have been hungrier, and the milks were in different containers (a bowl and a saucer), possibly in different amounts. It was also done only once. To improve it: give each cat both milks at the same time, side by side, in identical bowls with exactly 50 mL each. After 30 minutes, measure how much of each milk is left with a measuring cylinder. Repeat this on five different days, swapping which side each bowl is on, and average the results.
Explain the difference between a valid experiment and a reliable one.
A valid experiment actually tests what it is meant to test: it is a fair test where only the independent variable changes and the others are controlled. A reliable experiment gives similar results each time it is repeated. They are different: if Justin repeated his unfair cat test every day and the same cat always drank its milk, his results would be reliable, but still not valid, because he never tested the cats' preference fairly.