As a follow up to yesterday’s post on probability samples, I will be going over non-probability samples. These methods don’t provide the same level of fair representation, but certainly serve their own purpose in the market research industry. These types of samples lend themselves to preliminary forms of research, removing the need for them to be generalized to the population as a whole.
1) Convenience Sampling. Just as it sounds, this sample method is developed for the convenience of the research process. The process involves conducting surveys/research in convenient or high traffic areas. This is commonly used and best conducted in areas that contain the population of the target audience.
Example: A large University is looking to learn more about their student’s satisfaction with the campus facilities. They work with an independent market research firm to develop a survey. The data is then collected by setting up station in the library, the student recreation center and the cafeteria – where they pass out the surveys.
2) Judgment Samples. This is another form of non-random sampling; the respondents are chosen based on the interviewer’s observations. The process is quite subjective in that the interviewer has to use their experience and knowledge to determine which people are representative of the desired sample.
Example: A retail center is looking to provide a new line of accessories for eye glasses. They want to conduct some preliminary research to understand the thoughts of their target audience. To conduct their research, they interview people outside of their store and select only respondents who are wearing eye glasses.
3) Quota Sampling. This sample method is very similar to stratified sampling. The method includes categorizing respondents by their attributes and characteristics. The quotas are used in an attempt to match respondent attributes to that of the target audience you are hoping to learn more about. The difference between quota sampling and stratified sampling is that the respondents are not chosen randomly for quota sampling.
Example: A retail chain is hoping to open multiple locations. They are looking to conduct interviews to test some new concept ideas they have for marketing their stores. They know that their target audience is mostly males from age 30 to 50 in addition to a smaller amount of males from age 20 to 30. They decide to setup in a location outside of a gym and use their judgment to pick and choose the respondents that meet their quota.
4) Referral Samples. Also known as the snowball sample, this sample is based on the referral of other respondents. Referral samples can be useful if you are trying to reach a very specific niche that is not easy to find through other means.
Example: A manufacturer of basket weaving kits is considering changing their product’s message and redesigning their offering. They are having a hard time finding people that they know actively weave baskets, so when they conduct their research, they ask their respondents to refer to them other people that they know weave baskets.
Even though the research resulting from non-probability samplings can be valuable, it should be taken with caution as the respondents are not chosen through random selection. It is best to use your judgment when deciding on the right approach and executing the research – some of these sampling methods can cause extreme bias if not all factors are taken into consideration.