Statistics: Sampling
Sampling is the process of selecting a smaller group, or sample, from a larger population so that data can be collected and conclusions drawn about the whole population without surveying everyone. A Level Statistics covers the difference between a census and a sample, sampling frames, and named sampling methods including simple random, systematic, stratified, quota and opportunity sampling.
Before you start
No prerequisites. You can start here.
Method
- Decide whether a census (every member of the population) or a sample (part of the population) is more appropriate, weighing accuracy against time, cost and practicality.
- Identify the sampling frame: the list of all members of the population from which the sample will be drawn, and check whether it is complete and up to date.
- For a simple random sample, assign each member of the frame a unique number and use a random number generator (or random number table) to select the required number, rejecting any numbers outside the range or already chosen.
- For a systematic sample of size n from a population of size N, calculate the sampling interval k = N/n (rounded down), choose a random starting point between 1 and k, then select every kth member after that.
- For a stratified sample, divide the population into strata (groups), then select from each stratum a number proportional to its size in the population: (stratum size / population size) x sample size.
- State the advantages and disadvantages of the method used, referring to bias, cost, time and whether every member has a known (or equal) chance of selection.
Common mistakes
MistakeStarting a systematic sample at the first member of the list every time.
InsteadWork out the interval k, then choose the starting member at random from the first k members. The random start is the one random choice in a systematic sample, so leaving it out makes the sample the same every time.
MistakeTaking the same number of people from every stratum because equal groups seem fair.
InsteadA stratified sample takes from each stratum in proportion to its size: (stratum size / population size) x sample size. Equal numbers from unequal strata over-represent the small ones.
MistakeBelieving that a very large sample must be unbiased.
InsteadA large sample can still be biased by how members are selected or by an incomplete or duplicated frame. Random selection from a complete frame with one record per member avoids these frame defects; increasing sample size does not repair them.
MistakeDescribing quota sampling as random because the interviewer picks people within each group.
InsteadIn quota sampling the interviewer decides who fills each quota, so members do not have a known chance of being chosen. It is a non-random method; say so and name interviewer choice as the source of possible bias.
Worked examples
Example 1
A college has 720 students: 300 in Year 12 and 420 in Year 13. The college wants a stratified sample of 60 students, by year group. (a) Find the number of Year 12 students needed in the sample. (b) Find the number of Year 13 students needed in the sample. (c) Describe how the college should select which Year 12 students to include.
- Fraction of the population in Year 12 = 300/720.
- Number of Year 12 students in the sample = 60 x (300/720) = 25.
- Number of Year 13 students in the sample = 60 x (420/720) = 35 (or 60 - 25 = 35).
- To select which Year 12 students to include, number all 300 Year 12 students and use simple random sampling (e.g. a random number generator) to choose 25 of them.
Example 2
A firm employs 1500 people at three sites. A stratified sample is taken by site. The sample includes 15 people from Site A, which has 250 employees. (a) Find the total size of the sample. (b) Site B has 400 employees. Find the number sampled from Site B. (c) Site C has the remaining employees. Find the number sampled from Site C.
- In a stratified sample every stratum uses the same sampling fraction. From Site A the fraction is 15/250 = 3/50.
- Total sample size = 1500 x 3/50 = 90.
- Site B: 400 x 3/50 = 24 people.
- Site C has 1500 - 250 - 400 = 850 employees, so the sample from Site C is 850 x 3/50 = 51 people.
- Check: 15 + 24 + 51 = 90, which matches the total found in the first step.
- Final answer: (a) 90 people, (b) 24 from Site B, (c) 51 from Site C.
Example 3
In a fictional allotment association, the target population is its six current plot holders, labelled A to F, and the aim is to describe their opinions about opening hours. An old booking list is supplied as the sampling frame. Its seven records, in order, name A, A, B, C, D, D, E; there are no other records. The supplied selection rule chooses one of these seven records with equal probability and asks the named holder for an opinion. Identify the population and frame, then explain whether this rule gives every current holder an equal chance of selection. Discuss the resulting selection bias without proposing a different sampling procedure.
- The target population is the six current holders A, B, C, D, E and F. It is the group about which the conclusion is intended, rather than the records on the old list.
- The supplied sampling frame consists of seven booking records representing only five distinct holders. A and D each appear twice, while F is missing entirely.
- Each record has probability 1/7 of being chosen. A and D each occupy two different records, so each has probability 2/7 of being selected; B, C and E each have probability 1/7, and F has probability 0.
- Equal chances for records therefore do not give equal chances for people. Duplicate records favour A and D relative to the singly listed holders, while the missing record excludes F. Random choice from this frame does not remove these unequal selection chances.
- This is a problem with selection, even before anyone replies. The frame over-represents A and D and omits F, so treating the selected opinion as representative of all current holders risks bias. The premises give no holders' opinions, so they cannot establish which opening-hours opinion would be favoured.
- Final answer: the population is all six current holders; the frame is the seven supplied records. Selection chances are 2/7 each for A and D, 1/7 each for B, C and E, and 0 for F. The duplicate and missing records cause unequal representation, but the direction of any distortion in opinions cannot be determined from the information given.
Practice questions
Q1[1 mark]State what is meant by a census.Show answer
Answer: A survey that collects data from every member of the population.
Q2[1 mark]A school has an up to date, numbered list of all 1200 students. State the name for this list, in the context of sampling.Show answer
Answer: The sampling frame.
Q3[2 marks]A researcher wants a systematic sample of 20 from a population of 340. Calculate the sampling interval, k.Show answer
Answer: k = 17 (340 divided by 20).
Q4[2 marks]A gym has 450 members: 180 are under 30 and 270 are 30 or over. Find the number of under-30 members needed for a stratified sample of size 50.Show answer
Answer: 20 (50 x 180/450).
Q5[2 marks]A radio show asks listeners to call in and give their opinion on a new policy. State the name of this sampling method, and give one way it could produce a biased sample.Show answer
Answer: Self-selecting (volunteer) sampling; only listeners who feel strongly enough to call in respond, so the sample may over-represent extreme opinions.
Q6[3 marks]A stable has 60 horses, numbered 00 to 59. Describe fully how a simple random sample of 5 horses could be selected using a random number generator, ensuring no horse is chosen twice.Show answer
Answer: Generate two-digit random numbers, reject any greater than 59 and any repeats, and continue until 5 different horses have been selected.
Exam-style questions
Written in the style of a A Level Maths exam paper, with a full mark scheme.
Exam tips
- Tie each advantage or disadvantage to the situation in the question (cost, time, a group that could be missed) rather than stating a general fact about the method.
- When a proportional share is not a whole number, round each share and check that the rounded shares still add up to the sample size before writing the answer.
- Identify the complete list the sample would be drawn from in the question, and name any group it leaves out, before judging whether a method is suitable.
Q1[4 marks]A company holds a complete computer record of the annual mileage driven by each of its 3000 company cars. (a) State what is meant by a census. (1) (b) Give one advantage of the company using a sample of 150 cars rather than a census. (1) (c) Give one disadvantage of using a sample rather than a census in this case. (2)Show mark scheme
Tick each line you got.
Nothing ticked yet - 4 available
Q2[5 marks]A supermarket had 640 customers pass through its tills last Saturday morning. The manager wants to select a systematic sample of 20 of these customers to ask about checkout waiting times. (a) Calculate the sampling interval, k, the manager should use. (2) (b) Describe fully how the manager should select the systematic sample of 20 customers. (2) (c) The manager considers only sampling customers between 9am and 10am to make the sample quicker to select. Explain why this could introduce bias. (1)Show mark scheme
Tick each line you got.
Nothing ticked yet - 5 available
Q3[6 marks]A large secondary school has 1400 students across Years 7 to 13, and the school office holds a complete, up to date list of every student with their year group. (a) Give one reason why taking a sample, rather than a census of all 1400 students, would be appropriate here. (1) (b) Recommend a suitable sampling method for the head teacher to use, and justify your choice with reference to the information given. (3) (c) State one limitation of the method you recommended in part (b). (2)Show mark scheme
Tick each line you got.
Nothing ticked yet - 6 available
See real past-paper questions on Statistics: Sampling, organised by topic 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 A Level Maths past papers: 4 questions across 4 papers, latest tagged June 2022
- OCR A Level Maths past papers: 3 questions across 3 papers, latest tagged June 2024
- WJEC A Level Maths past papers: 3 questions across 3 papers, latest tagged Summer 2024
- OCR AS Level Maths past papers: 2 questions across 2 papers, latest tagged June 2024
Free printable worksheet
Want more practice on paper? Download the Statistics: Sampling worksheet pack - 14 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 1 of A Level Maths Statistics Essentials, the whole course as one free printable PDF. Its worksheet is also chapter 1 of A Level Maths Statistics Topic Workbook and chapter 1 of A Level Maths 10-Minute Tests 2, free printable practice books.
Other versions of this worksheet: