Non-Probability Sampling Methods

Survey research is still evolving within the scientific community. Many companies and universities have their own thoughts on survey research methodology. The International Institute for Educational Planning/UNESCO released their own document regarding how to design surveys for educational research.

The document addresses several aspects of survey research methodology, but focuses primarily on sampling – a serious roadblock for those studying students in educational institutions. The author starts off the article by reviewing the three most common types of non-probability sampling in educational research.

Judgment Sometimes it is hard to get a sample that accurately represents the subjects. So judgment sampling occurs when the researcher creates the sample themselves, looking for what the researcher believes is the best representation of the population. The greatest problem with judgment sampling is that researchers tend to disagree with what constitutes an accurate sample, leading to differing opinions about which research findings are “more correct” – an argument that is invalid in probability sampling.

Convenience Convenience samples are the most well-known type of non-probability sampling. Convenience sampling is simply sampling those around you, or those that you know you can reach. Polling a group of people that happen to be in a room would be a form of convenience sampling.

The main issue with this style of sampling is that it ignores everyone not in proximity. Quota Quota sampling has some similarities to judgment sampling, in that the researcher decides on the quota that “best represents the population.” Quota sampling takes the population’s known statistics, chooses what portion of the population is a large enough sample, and then creates a proportionate sample for their research. The greatest problem with quota sampling is not necessarily the quota itself, but the method used to choose those that enter the sample.

They also may not account for unknown aspects of the demographic.

Researchers agree that probability sampling is far more likely to yield accurate results. It is the best way to avoid any researcher error, and helps to guarantee that all of the data collected by the researcher is indisputable (even if the results themselves may not be accurate). Yet some researchers still prefer non-random sampling, or they may simply be limited in what they can do.

When that occurs, the above list represents most commonly used probability types that may be used in educational research. In the next post we’ll look at the types of probability sampling.