Research Tips

Probability versus Non-Probability Samples

Unlike a census, which attempts to capture data on all members of a given population, most market research relies on data collected from a subset or sample of the population and uses these to make inferences about the entire population. There are many different types of samples, but all of these can be classified as either probability or non-probability samples.

To be a probability sample, every member of the population has to have a known, non-zero chance of being included. Very few market research studies fully satisfy this criterion: for instance, people who don’t have telephones have a zero chance of being included in a telephone survey, and the probability of inclusion in many types of surveys is unknown.

Statistical measures such as confidence intervals and margins of error were developed for use with probability samples. Therefore, caution needs to be exercised when using them with the non-probability samples that are common in market research studies. To ensure reliable results, great care should be taken in the sampling process, even when it’s not possible to generate a true probability sample.