Want to keep learning?

This content is taken from the The University of Sheffield's online course, Making Sense of Data in the Media. Join the course to learn more.

Drawing a sample: non-probability sampling

In the last step, we learned how to draw a sample randomly by using probability sampling. Here, we’ll look at non-probability sampling, where samples are selected based on the subjective judgement of the researcher, rather than random selection.

In non-probability sampling, the chance of being selected is not known, and some units may not be given any chance to be selected. To draw a sample in a non-random way, you could select the first 500 people on a list, for example, or only those that live near you.

When you don’t have a good sampling frame, or when your resources are limited, non-probability sampling could be a better, or even the only, solution.

Quota sampling

The most popular non-probability sampling is quota sampling. This is where specific shares of interviews for the most important demographic characteristics are established beforehand. Researchers decide how many women and men, people of different ages, or different ethnicities, they want to interview.

In practice, research companies will mix random and non-probability sampling techniques by employing random root quota sampling. So they will start by drawing a random list of addresses, and then interview people according to the quotas that they have to fill. But they may skip addresses on their list to prevent high clustering of respondents in one neighbourhood. This type of sampling method is cheaper and less time consuming than probability random sampling.

Other types of non-probability sampling

  • Convenience sampling where the most accessible members of the target population respond to the survey on a first-come, first-serve basis. For example, through street recruitment.
  • Snowball sampling, where respondents invite their friends and family to participate in the survey.

These sampling methods do not produce information which is representative of the population, because some sub-populations could have been excluded from the sample. This means the results can’t be generalised to the population as a whole.

However, these methods are good for difficult to reach groups and for research on new and sensitive topics, such as drug-use, addictions, or undocumented migration.

In sum, whether to use probability or non-probability sampling depends on who your target population is, whether you have a good sampling frame for those people or institutions, and, as is often the case, how much time and money you have.

Share this article:

This article is from the free online course:

Making Sense of Data in the Media

The University of Sheffield