Experimental Design
Experimental design is the process of planning a scientific or statistical experiment so that it produces fair, reliable data and isolates the effect of the variable being tested. In GCSE Statistics this covers control groups, randomisation and reducing bias, so that any difference in results can be confidently linked to the variable under investigation rather than to chance or an outside factor.
Before you start
Make sure you're comfortable with these topics first:
Method
- Identify the independent variable (the one you change) and the dependent variable (the one you measure).
- Identify control variables that must be kept the same for every trial, so they cannot unfairly affect the result.
- Use a control group that does not receive the treatment, so it can be compared with the group that does.
- Assign participants or items to groups at random, to avoid bias in who ends up in each group.
- Use as large a sample as is practical, and repeat trials to check the results are reliable.
- Consider using a blind or double-blind procedure where relevant, so participants or researchers cannot influence the result.
Common mistakes
MistakeChanging more than one thing between the groups, for example giving one group a new revision app and also extra lesson time, so nobody can say which change caused any difference.
InsteadChange only the independent variable and hold everything else the same for both groups. Anything else that differs is a control variable that has not been controlled.
MistakeThinking that random selection and random allocation are the same, so a random sample is chosen and then people are placed in groups by choice.
InsteadRandom allocation means chance decides which group each person joins, for example by drawing names or using random numbers, even after the sample has been chosen.
MistakeGiving the control group nothing at all, so one group takes part in an activity and the other sits idle, and the difference may come from attention or expectation rather than the treatment.
InsteadGive the control group the same routine, time and attention, replacing the treatment with a neutral alternative such as a placebo where possible, so that the treatment is the only difference.
Worked examples
Example 1
A company wants to test whether a new fertiliser increases tomato plant height. Describe a fair experimental design, including a control group.
- Identify the variables: independent variable is whether fertiliser is used, dependent variable is plant height after a fixed time.
- Identify control variables to keep constant: same plant species, same amount of water, same amount of sunlight, same size of pot.
- Take a large sample of plants and randomly assign half to the fertiliser group and half to a control group with no fertiliser.
- Grow all plants under the same conditions for the same length of time, changing only the fertiliser.
- Measure the height of every plant at the end and compare the mean height of the two groups.
Example 2
Twelve students were randomly split into two groups of six to test whether a new type of running shoe improves sprint time. Both groups ran the same 60 m course on the same track, one at a time. Times in seconds. New shoes: 9.8, 10.4, 9.9, 10.1, 10.6, 9.7. Standard shoes: 10.2, 9.9, 10.5, 10.0, 10.3, 10.7. (a) Find the mean time for each group. (b) Does the experiment show that the new shoes improve sprint time? Give two reasons and one improvement.
- New shoes: the total is 9.8 + 10.4 + 9.9 + 10.1 + 10.6 + 9.7 = 60.5, so the mean is 60.5 / 6 = 10.08 seconds (2 d.p.).
- Standard shoes: the total is 10.2 + 9.9 + 10.5 + 10.0 + 10.3 + 10.7 = 61.6, so the mean is 61.6 / 6 = 10.27 seconds (2 d.p.).
- The new-shoe mean is lower by about 0.18 seconds (from the unrounded means), and the design is reasonably fair: random groups, one course, one track.
- Reason 1: the difference in means is small compared with the spread. The new-shoe times range over 0.9 seconds and the standard times over 0.8 seconds, and the slowest new-shoe time, 10.6, is slower than five of the six standard times.
- Reason 2: with only six runners in each group, chance differences in the runners' ability could easily produce a gap this size.
- Improvement: use many more runners in each group, or repeat the trial, and compare the means again.
- Final answer: the new shoes' mean is 10.08 s against 10.27 s, but the data do not convincingly show an improvement because the difference is small, the ranges overlap and the groups are small.
Practice questions
Q1[1 mark]What is meant by the independent variable in an experiment?Show answer
Answer: The variable that the experimenter deliberately changes or controls.
Q2[1 mark]What is a control group used for?Show answer
Answer: To give a baseline for comparison, showing what happens without the treatment being tested.
Q3[2 marks]Give one reason why participants should be assigned to groups at random.Show answer
Answer: To avoid bias, so pre-existing differences between participants are spread evenly between the groups.
Q4[2 marks]A scientist tests a new plant food but only uses it on the healthiest-looking plants. Explain why this is a flawed design.Show answer
Answer: The sample is not randomly assigned, so results cannot be fairly compared with a typical control group.
Q5[1 mark]What is meant by a control variable?Show answer
Answer: A variable that is kept the same for every group or trial, so it cannot affect the result.
Q6[2 marks]In a medicine trial, half the patients receive the real drug and half receive a dummy pill, and neither the patients nor the doctors know who received which until after results are recorded. What is this type of trial called, and why is it used?Show answer
Answer: A double-blind trial; it prevents the patients' or doctors' expectations from biasing the results.
Exam-style questions
Written in the style of a GCSE Statistics exam paper, with a full mark scheme.
Exam tips
- Name the independent variable, the dependent variable and how the dependent variable will be measured, with units, before you list anything else; a plan that cannot be measured cannot be tested.
- Give each control variable a specific setting (the same time of day, the same equipment, the same amount of time) rather than writing 'keep it fair'.
- When asked to improve a design, say what problem your change removes as well as what you would do differently.
Q1[2 marks]Explain why a sample size of 4 plants is not sufficient to draw a reliable conclusion in an experiment.Show mark scheme
Tick each line you got.
Nothing ticked yet - 2 available
Q2[3 marks]A gardener wants to test whether talking to plants increases their growth. Describe a fair test, including how a control group should be used.Show mark scheme
Tick each line you got.
Nothing ticked yet - 3 available
Q3[4 marks]A researcher is investigating whether a new teaching method improves test scores. She lets students choose which method they use. (a) Explain one problem with this design. (b) Describe two improvements that would make the experiment fairer.Show mark scheme
Tick each line you got.
Nothing ticked yet - 4 available
See real GCSE Statistics past-paper questions, with official mark schemes
This topic in real past papers
We have tagged some boards' past papers question by question (not every board or year yet). This topic comes up in:
- AQA GCSE Statistics past papers: 13 questions across 11 papers, latest tagged June 2023
- Edexcel GCSE Statistics past papers: 4 questions across 4 papers, latest tagged June 2022
Free printable worksheet
Want more practice on paper? Download the Experimental Design worksheet pack - 31 pages of exam-style questions with a full mark scheme. Free to download, no card or account. Print it for personal and classroom use.
This topic is chapter 10 of GCSE Statistics Foundation Essentials 1 and chapter 10 of GCSE Statistics Higher Essentials 1, the whole course as one free printable PDF. Its worksheet is also chapter 2 of GCSE Statistics Foundation Topic Workbook 2 and chapter 3 of GCSE Statistics Higher Topic Workbook 2, free printable practice books.
Other versions of this worksheet: