Population, sample and sampling techniques
Studying this course? Keep your progress in one place.
Key facts
- The population is everything of interest; a sample is a selection from it; a census observes every member.
- A sample gives informal inferences about the population; a census is the most accurate but costly or impossible.
- In a simple random sample every member has an equal chance. Needs a sampling frame; ignore repeats.
- Stratified: sample each stratum in proportion to its size. Systematic: every th member, .
- Quota and opportunity sampling are not random, so are likely to be biased.
Population and sample
A census observes the whole population; a sample observes part of it.
The population is the whole set of items or people of interest and a sample is a selection from it. A census observes every member.
We use a sample to make informal inferences, such as estimating a mean or proportion. The list of all members is the sampling frame.
- 1
Population
the whole set of interest; a census observes all of it
- 2
Sampling frame
the list of every member
- 3
Sample
a selection from the frame
- 4
Inference
informal conclusions about the population
| Census | Sample | |
|---|---|---|
| Accuracy | Gives the true value | Only an estimate, may not be representative |
| Cost and time | Expensive and slow | Cheaper and quicker |
| Practicality | Sometimes impossible, e.g. crash-testing every car | Often the only option |
Census
- Accuracy:
- Gives the true value
- Cost and time:
- Expensive and slow
- Practicality:
- Sometimes impossible, e.g. crash-testing every car
Sample
- Accuracy:
- Only an estimate, may not be representative
- Cost and time:
- Cheaper and quicker
- Practicality:
- Often the only option
Why is a sample used to test the lifetime of light bulbs?
Simple random sampling
Number the sampling frame, generate random numbers, ignore repeats, and select those members.
In a simple random sample of size , every possible sample of size has an equal chance, so every member does too.
Advantages: free from researcher bias. Disadvantages: needs a complete sampling frame, may be impractical for a large or spread-out population, and can miss a small group by chance.
- 1
Number the frame
Label each member 1 to N.
- 2
Generate numbers
Produce n random numbers in that range.
- 3
Ignore repeats
Discard any number already chosen and generate another.
- 4
Select members
Take the members with those numbers.
A school has 800 students on its register. Which describes a correct first step for a random sample of 40?
Systematic and stratified sampling
Systematic takes every th member after a random start; stratified samples each group in proportion to its size.
A systematic sample takes a random start, then every th member of an ordered list, . It is quick, but biased if the list has a matching pattern.
A stratified sample divides the population into strata (such as year groups) and takes a random sample from each in proportion to its size. It is representative when there are clear groups, but needs information about them.
- Number from stratum
Worked example
A school has 900 students: 360 in Year 10, 300 in Year 11 and 240 in Year 12. Find a stratified sample of 60.
- 1
Year 10: .
- 2
Year 11: .
- 3
Year 12: .
- 4
Check the total: .
A list has names and a systematic sample of is needed. What is the interval ?
Quota and opportunity sampling
Quota and opportunity sampling choose people by convenience, so they are not random and may be biased.
In quota sampling an interviewer fills a quota for each group (for example 20 men and 20 women), often by convenience. In opportunity sampling you take whoever is available and willing.
Advantages: quick, cheap, no sampling frame needed. Disadvantages: not random so likely to be biased, and there is no way to measure the likely error.
- 1
Set quotas
for example 20 men and 20 women
- 2
Choose by convenience
the interviewer picks whoever is available
- 3
Stop when each quota is full
no sampling frame is needed
- 4
Possible bias
the sample is not random
Random
- Simple random
- Systematic
- Stratified
Not random
- Quota
- Opportunity
- Interviewer chooses
An interviewer must survey 20 men and 20 women and stops people in the street until both quotas are full. Which method is this?
Selecting and critiquing a method
Match the method to the population, then comment in context on who might be missed.
Consider whether there is a sampling frame, whether the population has distinct groups, the cost and time, and the accuracy needed.
Different samples give different results, called sampling variability, so conclusions are informal inferences with uncertainty. Unrepresentative samples show bias. A larger sample generally varies less.
- 1
Is there a sampling frame?
if so, random or systematic sampling is possible
- 2
Are there distinct groups?
if so, consider stratified sampling
- 3
Consider cost, time and accuracy
quota and opportunity are quick and cheap
- 4
Comment in context
say who might be missed and so where bias could arise
Method
- Simple random
- Systematic
- Stratified
- Quota or opportunity
Best when
- There is a full sampling frame
- The list is ordered
- There are clear groups
- No frame, quick and cheap
Two different random samples from the same population give different means. Why?
Try an exam question
A company has 480 employees: 192 in sales, 168 in production and 120 in administration. A stratified sample of 40 is to be taken. Find the number from each department and describe how the sales employees would be chosen.
[4 marks]
- [1]Sales: .
- [1]Production: .
- [1]Administration: .
- [1]Number the sales employees 1 to 192 and select 16 using random numbers, ignoring repeats.
That's the notes covered.
Carry on to the next subtopic.
Try a question
Which of the following is the population in this investigation?
1 mark, marked against the mark scheme
Which of the following is the population in this investigation?
1 mark, marked against the mark scheme