A larger sample reduces some uncertainty, but cannot repair a biased selection method.
The population is the full set you want to understand. A sample is the smaller set you observe. Your sampling frame is the list or set from which you select. A list of community-centre visitors excludes people who rarely leave home. Be explicit about that limit.
Random sampling gives each unit in the frame a known chance of selection. Systematic sampling selects every nth unit after a starting point; a repeated spatial pattern can distort it. Stratified sampling divides the population into meaningful groups, then samples within them. Consider age bands, housing areas or distance zones when these may affect the results.
Convenience sampling selects whoever is easiest to reach. It saves time but can miss working residents, people with limited mobility or people using a different language. Purposive interviews deliberately include people with relevant experience; they provide depth but do not estimate population percentages by themselves.
Choose sample size with the variability, subgroup comparisons and resources in mind. Include different days and times where experience changes. Record refusals and missing groups. Increasing 20 convenient interviews to 100 convenient interviews does not make the selection representative.
Step by step
Define the frame
List the streets, households, sites or times eligible for selection.
Choose and justify
Explain how the method includes relevant variation without selecting only expected answers.
Report limits
State subgroup counts, timing and non-response so readers can judge the reach of your findings.
Worked example: A workable improvement
A survey at one bench at 10 am mainly reaches people already using the park. Add selected residential blocks, different time slots and an accessible interview option. This improves inclusion more directly than asking 50 extra people at the same bench.
Watch out for this
Equal numbers from each group always represent the population.
Equal numbers help compare groups. Estimating a population total may require proportional sampling or justified weighting.
Check your understanding
A weekday lunchtime survey misses many employed residents. Which improvement targets this problem?
- Ask twice as many people at lunchtime.
- Use a longer questionnaire.
- Add evening and weekend sampling periods.