What this lesson teaches
I can choose a sampling method and collect primary data with questionnaires, interviews and mental maps.
Syllabus K229, Topic 1.3, Key Question 2. How to collect primary data?
I can judge the reliability and validity of an investigation and say how to improve it.
Syllabus K229, Topic 1.3, Key Question 3 (part). How to process and analyse data?
Make a guess
Students survey 300 people at one park entrance to find out how all residents use the park. Is the large sample enough to make the result reliable?
- Yes, as long as they ask at different times of day.
- No. A survey cannot tell us about how a park is used.
- Yes. 300 people is a large number.
- No. Everyone asked already uses the park.
Show the answer
No. Everyone asked already uses the park.
The sample misses residents who never visit. Asking more people at the same entrance repeats the same bias instead of fixing it.
How you choose your sample decides whose experiences your results can represent.
A sample is the people or sites you choose to study. The target population is the whole group your question is about. A sampling frame is the list you choose from. If your question is about all residents, a list of park visitors will miss people who never go to the park.
Convenience sampling picks whoever is easiest to reach, such as people walking past one entrance. Quota sampling sets a number for each group, then picks people non-randomly within it. Both are practical, but where you stand or whom you choose can favour some experiences over others. This is sampling bias.
Simple random sampling uses a random process, so everyone on the list has an equal chance of being picked. Stratified random sampling first splits the population into relevant groups, then picks randomly within each group.
Suppose 60% of residents live near the park and 40% farther away. A proportional stratified sample of 50 could include 30 nearby and 20 farther-away residents. Within both groups, people must still be picked randomly. Filling those numbers with whoever is easiest to reach would be quota sampling instead.
Every method has limits. The list may miss a group, some people may not reply, or answers may be inaccurate. Surveying at one time may miss some users. Asking more people at the same entrance just repeats the same bias. Match the fix to the problem: to include non-users, use a list of all residents.
- Convenience
- Whoever is easiest to reach.
- Quota
- Set numbers per group, chosen non-randomly.
- Stratified random
- Split into groups, then choose randomly in each.
Step by step
Define the population
Who is the question about?
Choose the method
Explain how it includes the right groups.
Note who is missing
Non-responses, times not covered and absent groups.
Worked example: Fix the right problem
A lunchtime survey at the park entrance misses workers and people who never visit. Surveying at another time helps with workers, but still misses non-users. A list of all residents fixes that second problem.
Watch out for this
Quota sampling is random because it includes different groups.
Quotas set how many from each group. Only random selection within groups makes it stratified random sampling.
Check your understanding
Students split a resident list into two distance groups and use a random number generator in each. What method is this?
- Convenience sampling by group.
- Stratified random sampling.
- Quota sampling.
Show the answer
Stratified random sampling.
Yes. Groups first, then random selection within each.