Stockertim Dictionary · Marketing
Representative sample
Also called representative survey sample
In short
A group of respondents chosen so that it resembles the wider population on the characteristics that matter to the study, such as age, region and buying habits.
- Intermediate
- Concept
- Checked
In detail
No survey can ask everyone, so it asks a sample and uses the answers to stand for the whole population. That works only if the sample looks like the population in the ways that affect the answers. A representative sample of bread buyers, for instance, has roughly the same mix of ages, regions and household sizes as all bread buyers.
The strongest approach is a probability sample, in which every member of the population has a known chance of being selected, for example from a list of addresses. It is the method official statisticians prefer and also the most expensive. Commercial research more often uses quota samples, recruiting until each group, such as younger women in one region, reaches its real share of the population.
After fieldwork, weighting corrects the remaining gaps. An under-represented group counts for a little more and an over-represented group for a little less, so that the weighted totals match known population figures such as census counts. Heavy weighting is a warning sign, because a few people end up speaking for many.
Convenience samples, made up of whoever is easiest to reach, such as visitors to a brand's own website, are quick but self-selected, and they tend to lean towards the most engaged customers. Size does not fix this: a very large sample drawn from the wrong people repeats the same distortion with more apparent confidence.
Worked example
A survey of 400 adults needs 30% of them to be in the youngest age band to match the population, but an online panel returns 180 people in that band.
- Quota for the youngest band: 400 × 30% = 120
- Weight for each younger respondent: 120 / 180 = 0.67
Each younger respondent counts as about two-thirds of a person, so the group's views do not outweigh its real share.
In practice
Opinion polls, official surveys and brand trackers publish how their samples were drawn and weighted. Buyers of research ask for this, along with the unweighted count behind each figure, before relying on a result.
Watch out
- Matching the population on age and region does not guarantee a match on the behaviour being studied.
- A self-selected online sample can be large and still unrepresentative.
Learn it properly
Official sources
- Uncertainty and how we measure it · Office for National Statistics
- Sample size and data quality · US Census Bureau
- Quality in official statistics · Office for National Statistics
Facts last checked on 3 November 2026.
Education, not advice: the Stockertim Dictionary explains what terms mean. It never recommends buying or selling anything.