Glossary
Representative Sample
What is a representative sample?
A representative sample adequately reflects a defined population for the characteristics and estimates of interest. Representativeness is not a single property guaranteed by sample size or matching a few demographics.
Why a representative sample matters
If the aim is to describe a population, the sample needs to support that inference. The question is not only who was studied, but which conclusions can reasonably extend to people outside the sample. This is part of assessing external validity.
How representative sampling works
Probability sampling uses known selection probabilities, which need not be equal. Common designs include:
- Simple random sampling: Units in the sampling frame have equal chances of inclusion.
- Stratified sampling: Divide the population into groups, then sample within each group. Allocation need not match population proportions; a small group can be oversampled when appropriate weighting is used for population estimates.
- Cluster sampling: Select groups, such as schools, and study all or a sample of units within them. Each selected cluster need not mirror the entire population.
The design and analysis must account for how participants were selected.
How is representativeness assessed?
Assess the target population, sampling frame, participation and relevant population comparisons. Weighting can address known design differences but cannot automatically repair every bias. A margin of error describes sampling uncertainty under stated assumptions, not overall representativeness or errors from nonresponse and measurement.
Representative sample example
A national opinion poll aims to estimate views in a defined population. Matching age groups can be useful, but does not show that respondents and nonrespondents have the same views. The selection process and who actually takes part matter alongside the final demographic totals.
Practical limitations
Costs, incomplete population lists, and nonresponse can restrict who is included. Population characteristics can also change over time. A useful report states these limits and defines the population and period its estimates concern.
Representative sampling vs random assignment
Sampling people from a population differs from randomly assigning study participants to treatments. Neither procedure by itself ensures that treatment findings apply to every population or setting.