Stratified sampling means you split your population into meaningful groups (strata) first — by level, gender, department, or faculty — then sample randomly from each group. This guarantees every important subgroup is represented, instead of leaving it to chance. It is the go-to method when your population is clearly made up of different groups you want to compare or balance.
When to Use It
- Your population has clear subgroups that matter to your study.
- You want each subgroup fairly represented, not left to luck.
- You plan to compare results across the subgroups.
- A simple random sample might miss or under-represent a smaller group.
When Not to Use It
- Your population is uniform with no meaningful subgroups.
- You cannot reliably sort people into the groups you would need.
- The added complexity is not worth it for a simple, small study.
How to Do It (Plainly)
- Decide the subgroups (strata) that matter — e.g. 100, 200, 300, 400 level.
- Work out how many of your sample should come from each, usually in proportion to their size.
- Randomly select that number from within each subgroup.
- Combine them — your sample now reflects every group.
Nigerian Project Example
Studying study habits across a faculty with 100–400 level students, you split the population by level, then randomly pick respondents from each level in proportion to its size — so no level is over- or under-represented in your final sample.
Undergraduate vs Postgraduate
Undergraduates use stratified sampling when their topic compares groups (such as male vs female, or by level). Postgraduates often combine stratification with larger samples and multi-stage designs, and are expected to justify exactly how the strata were defined and proportions calculated.
Common Mistakes
- Defining strata that have nothing to do with the research questions.
- Taking an equal number from each group when the groups are very different sizes (unless you intend disproportionate sampling and explain it).
- Forgetting to sample randomly within each stratum.
State your strata and proportions in your methodology — see how to write Chapter Three and the full research methodology guide. Compare with simple random sampling and systematic sampling. Project Lab can help you set fair proportions.
Frequently Asked Questions
What is a stratum?
A stratum is one subgroup of your population that shares a characteristic — for example, all 200-level students. Stratified sampling makes sure each stratum is represented in the sample.
How do I decide how many to take from each group?
Usually in proportion to each group's share of the population. If a group is 30% of the population, about 30% of your sample comes from it. Explain your proportions in your methodology.