Population and sample; representativeness and bias; sampling techniques
Machine translation
Learn
To find out about a large group, you cannot always ask everyone. The population is the whole group you want to study. It is made up of individuals: people, but also bread rolls, packages or batteries.
The sample is the part of the population that you do ask or measure. Studying the whole population is called a census. You use a sample when the population is very large, when asking everyone takes too much time, or when measuring destroys what you measure, such as testing how long a battery lasts. The population size is written N and the sample size is written n.
Population: N = 100 individuals. Sample: the 10 marked, n = 10 A sample is useful if it is representative: if it is like the population in what you are studying. If it is not, it has bias: its answers almost always lean the same way. Size does not fix it: a large sample that is badly chosen is still biased.
Bias comes from choosing the individuals nearest to you, from letting only volunteers answer, from drawing lots from an incomplete list, or from asking a question that pushes the answer.
Asking only the football team gives a biased answer To avoid bias, you choose the sample at random. In simple random sampling, every individual has the same chance: you draw lots. In systematic sampling, you put the population in order, draw lots for the first one and then take one in every k, where k = N ÷ n. In stratified sampling, you split the population into different groups, the strata, and in each one you draw lots for a part in proportion to its size.
Bread Bread rolls Sample Bars 240 24 Ring doughnuts 120 12 Plaits 40 4 Total 400 40 Stratified sample: 40 out of 400, one tenth from each stratum With a well-chosen sample you can estimate the population. If 20% of the sample prefer dancing, you expect a similar answer in the population, but not an identical one: another sample would give another number.
Worked examples
Example 1
Vera has a numbered list of 40 visitors and wants to ask 8, using systematic sampling. She draws lots for the first one between 1 and 5 and gets 3. Which visitors does she ask?
- Work out the step: k = N ÷ n = 40 ÷ 8 = 5. Take one in every 5.
- Start at 3, the first one drawn between 1 and 5.
- Add 5 each time: 3, 8, 13, 18, 23, 28, 33 and 38.
- Check that there are 8, the sample size.
Visitors 3, 8, 13, 18, 23, 28, 33 and 38.
Watch out! Here many people always start at 1, instead of drawing lots for the first one between 1 and 5.
Example 2
Vera has 150 visitors: 90 children, 45 young people and 15 adults. She wants a stratified sample of 30. How many does she choose from each group?
| Group | Total | Sample |
|---|---|---|
| Children | 90 | ? |
| Young people | 45 | ? |
| Adults | 15 | ? |
| Total | 150 | 30 |
- The sample is 30 out of 150: one fifth, .
- Each stratum gives one fifth of its size: 90 ÷ 5 = 18, 45 ÷ 5 = 9 and 15 ÷ 5 = 3.
- Check: 18 + 9 + 3 = 30, the sample size.
- Within each stratum, you draw lots for who is asked.
18 children, 9 young people and 3 adults.
Watch out! Here many people choose 10 from each group, and then the adults count as much as the children.
Example 3
Tony wants to know which sport the 600 pupils at his school prefer. He asks the 20 pupils in his football team. What is wrong and how can he fix it?
| Data | In Tony's study |
|---|---|
| Population | The 600 pupils |
| Sample | The 20 in the team |
| Variable | Favourite sport |
| How they were chosen | The ones nearest to him |
- The population is the 600 pupils at the school. The sample is the 20 in the team.
- The team choose football more than everyone else: the sample is biased, because Tony chose the ones nearest to him.
- To fix it, you have to choose at random: number the 600 pupils and draw lots for 20 numbers (simple random sampling).
- This way every pupil has the same chance of being chosen.
The sample is biased. With a draw among the 600 pupils, everyone has the same chance.
Watch out! Here many people think it can be fixed by asking more pupils in the football team.
A challenge to think about
Design a survey to find out which sport the pupils at your school prefer. Write who the population is, how many individuals your sample would have and how you would choose it. Say which technique you would use, why, and which biases you would have to avoid.
Practise
Rounds of 10 questions. Choose a level:
Your grade in this topic:No grade yetDo a round at the Basic level to get a grade.