TL;DR:
A fair test changes one thing, the independent variable, measures one outcome, the dependent variable, and keeps every other factor the same.
A control group is the comparison that shows what happens without the change.
Repeated trials and a mean help you tell a real effect from chance.
A student claims that bean plants grow taller in more light. She puts one plant on a sunny window and another in a dark cupboard, waters one more than the other, and a week later announces that light works. What is wrong with that experiment? Almost everything. Fixing it teaches you what a controlled experiment is.
The idea in plain words
An experiment asks whether changing one thing causes a change in another. To trust the answer you need to make sure nothing else changed. The structure is simple.
Independent variable: the one factor you deliberately change (hours of light per day).
Dependent variable: what you measure in response (plant height).
Controlled variables: everything else you keep the same (water, soil, pot size, temperature, seed type).
Control group: the set of plants kept at normal conditions, so you have something to compare with.
Repeated trials: several plants per condition, not just one.
Why each part is there
If two things change at once, you cannot tell which one caused the effect. Controlled variables make the test fair. A control group gives a baseline. And because individual plants vary, one plant per group might just be a strong or weak seed. Several plants let you average that out.
Worked example: bean plant growth under different light
Question: Does the number of hours of light per day change how tall bean seedlings grow in two weeks?
State a testable prediction: Seedlings given more hours of light per day will grow taller after 14 days.
Choose the variables: Independent: light, at 4, 8 and 12 hours per day. Dependent: height in cm after 14 days.
Control everything else: Same bean variety, same soil, same pot size, same amount of water each day, same temperature, same light source and distance.
Set up groups: Three groups of 5 plants each. The 12 hour group is the control, since it matches the normal day length in the lab.
Measure: Measure height from the soil to the tallest tip with a ruler, at the same time of day.
Analyse: Work out the mean for each group and look at the spread.
Here are practice results, invented for the exercise. Heights in cm after 14 days.
Light per day | Heights of the 5 plants (cm) | Sum (cm) | Mean (cm) |
|---|---|---|---|
4 h | 5, 6, 4, 5, 5 | 25 | 5.0 |
8 h | 8, 9, 10, 9, 9 | 45 | 9.0 |
12 h (control) | 12, 13, 11, 12, 12 | 60 | 12.0 |
The means rise with light: 5.0, 9.0 and 12.0 cm. The ranges within each group (6 − 4 = 2, 10 − 8 = 2, 13 − 11 = 2) are small compared with the differences between groups (about 4 and 3 cm), so the pattern is unlikely to be chance. The data supports the prediction for these conditions.
Fixing the student's first attempt
Go back to the first experiment and list what was wrong. Light and water both differed, so the cause was unclear. There was one plant per condition, so chance could explain the result. A dark cupboard also changes temperature, and bean seedlings in the dark grow in a different way, so the test was not fair even about light alone. And no one said how height would be measured.
The redesigned version fixes each point: one independent variable, controlled conditions, three light levels, five plants per level and a fixed method of measuring. This is a general pattern. Take a bad design, find each place where two things differ or where there is no comparison, and fix them one at a time.
Limits of what it shows
It covers 4 to 12 hours only. Do not claim anything about 20 hours, or constant light.
Height is one measure of growth. Mass or leaf number could tell a different story.
Light hours is not the same as light intensity or colour. Those were not varied here.
A fair test does not need a statistical formula to be useful, but a real study would use a statistical test as well.
How to check your plan
Can you name the single independent variable?
Is every other factor on your controlled list?
Is there a control group, and do you say why it is the control?
Are there enough repeats to see the spread?
Can someone else follow your steps and get a similar result?
Common mistakes
Changing two things at once, such as light and water.
Using one plant per group and treating its height as the result.
Calling something the control group without saying what it is compared with.
Measuring in different ways for different groups, such as to the highest leaf for one and the stem for another.
Concluding more than the test can show.
Where you will use it next
The same design works for enzyme experiments, osmosis in potato slices, reaction times and any lab where you ask whether one factor affects another. To practise reading the results, see reading a graph from a biology experiment, and for exam-style design questions, AP Biology experimental design questions.
Frequently asked questions
Is a control group the same as controlled variables?
No. Controlled variables are the factors you keep constant. The control group is the comparison group that does not get the treatment, or gets the usual condition.
Why do we repeat trials?
Individual results vary. Repeats show how big that variation is and let you average it.
Get started
