Sampling Techniques and Who Takes Part

Everything in this topic ends up here. However well designed a study is, it can only tell you about the people who were in it — and how those people were chosen decides how far the findings reach. Five techniques, each with a clear trade between how easy it is and how much you can claim afterwards.

Start with the population

Before choosing a technique, ask who the findings are supposed to be about. A researcher studying the experience of being a single teenage parent must draw from that specific group, because nobody else has that experience. That is a distinct target population.

Compare that with a researcher testing the duration of short-term memory. There is no reason to think memory works differently for a plumber than a pharmacist, so any adult across a wide age range would do. No distinct population is required, and the sampling can be much broader.

Four common techniques, side by side

Opportunity sampling

Also called convenience sampling. The researcher takes whoever is available and willing: shoppers at eleven in the morning, students in a lecture, parents at a baby group.

Self-selecting sampling

Also called volunteer sampling. The researcher advertises — posters on a campus, a social media post, a newspaper advert — saying when and where the study is happening and how to take part. Adverts can also ask for specific characteristics, such as first-time parents or bilingual speakers.

It is quick, cheap and probably the most used method in psychology. Participants have made a conscious choice to be there, so they tend to be willing and enthusiastic, and are less likely to jeopardise the study.

Random sampling

Every member of the target population has an equal chance of being selected. In practice that means putting every name into a container and drawing until the sample size is reached, or using name-generator software when the sample needs to be large.

• Strength: it removes researcher bias entirely, because the researcher has no control over who is chosen.
• Strength: the sample should be fairly representative, so findings can be generalised to the target population.
• Limitation: it is time-consuming and often impractical, since you rarely have a complete list of the population.
• Limitation: not everyone selected will agree to take part, which quietly undoes the randomness.
• Limitation: chance can still produce an unbalanced sample, such as one that happens to be all male.

Stratified sampling

Stratified sampling builds a small-scale reproduction of the target population. The population is divided into categories that matter to the research — age, gender, education level, ethnicity, profession — and each category contributes to the sample in proportion to its share of the whole.

If 18 per cent of the population are men aged 30 to 40, then 18 per cent of the sample will be too. That proportionality is what makes it the strongest method for generalisability, and it also lets the researcher control which categories are used, choosing the ones relevant to the aim.

The costs are practical. Researchers cannot always confidently classify every member of a population into a subgroup, and gathering the sample takes time, since you need enough information about the population to divide it up in the first place.

Snowball sampling

Some populations are almost impossible to reach through an advert. People may belong to a group that is closed or hard to access, may not respond to conventional recruitment, or may feel nervous or compromised if a researcher approached them directly.

Snowball sampling solves this by starting small. The researcher finds one or a few participants and asks them to recruit others in a similar situation. Those new participants recruit more, and the sample grows outward.

The strengths are real: hard-to-reach populations get represented in research at all, and being introduced by someone in a similar situation instils trust, which matters enormously in qualitative work where rapport shapes what people are willing to say.

The limitations follow from the same mechanism. Participants recruited by each other are likely to report similar experiences, so the scope of the research narrows. The researcher also has little control over who joins, relying on recommendations and referrals from other people, which can threaten the credibility of the findings.

Worked examples