Snowball Sampling

Snowball Sampling is a data or sample-gathering technique in which the current participants help the researcher identify and add new potential participants or subjects. The process keeps on going until the researcher finally reaches a saturation point. It is a non-probability method.

The snowball sampling method is mostly used in qualitative research in which the population is hard to seek or identify. The process starts with a small group of initial participants that work as data seeds that reach out to other potential participants they know in the target population; the technique is useful when studying sensitive topics.

• Snowball sampling is a recruitment technique in which initial units help in gathering, identifying and gathering future potential participants for the study.
• The technique has three main types: linear, exponential discriminative, and exponential non-discriminative. The snowball sampling is also referred to as chain or network sampling.
• The common application of this method is phenomenology research with sensitive topics, which is open to many biases.
• There is a bias problem present in this method, which proper sample selection can eliminate.

Snowball sampling is a data recruitment technique in which the initial sample population refers to the future potential samples for study. It works like a network or a chain of people or subjects contacting others with similar traits, aspects or characteristics required for the study and bringing them in as the new sample for the researcher. With such a sampling method, the researcher can add more subjects to the study for analysis until they reach the ideal sample size.

It most commonly finds application in phenomenology research where the subject is hidden and does not want to come out or reveal its identity. The technique saves time and effort for the researcher because the initial sample takes on the responsibility of networking and brings new subjects to the researcher. Whenever there is a sensitive topic or a social survey about an uncommon subject, the snowball sampling method proves to be very beneficial.

The key limitation or problem with snowball sampling is that when initial samples refer to the new sample, there is a degree of bias in sample subjects; researchers must run tests on the new referrals to ensure they are fit for the study. One can do this through different types of snowball sampling that allow a researcher to properly recruit new samples for the study by eliminating biases.

The types of snowball sampling are -

1. Linear snowball sampling - This type of sampling follows a straight line of referral; it starts with one initial subject that provides one new referral, which further recruits into the sample group till the completion of the ideal sample size.
2. Exponential non-discriminative sampling - In this technique, the first subject recruited as the initial sample provides a referral of multiple potential samples, who, when recruited, go on offering new referrals individually to the researcher to the extent till the desired sample size is achieved.
3. Exponential discriminative sampling - This is the third type of snowball sampling, which is similar to exponential non-discriminative snowball sampling, as each subject offers more than one referral. Still, the distinctive factor here is that in this type of technique, there is recruitment of only one referral from multiple referrals. The researcher makes this decision based on the objective of the study.

Below are two examples of snowball sampling -

Example #1

Suppose a doctor visits a village and is researching a very rare disease. He comes across four people who believe they have the same disease. Since it is a village, many people have the same disease but also have notions and stigmas about the disease and the new doctor and do not wish to come forward and identify themselves.

In this case, the doctor who has the initial sample of four people asks them to help him in his research. The four people go out and connect with other people they know as neighbors, friends, relatives and colleagues to clear their doubts and notions and bring them to the doctor as a new sample for the disease study. It is a straightforward example of snowball sampling. The sample was present but hidden, and the initial sample helped in bringing in potential new samples to the study.

Example #2

Another hypothetical example could be YouTube video recommendations. Suppose an individual is searching for a particular food dish video and starts watching the first five in the search results. In every video surfing window, there will be a recommendation section that will refer the viewer to similar food dish videos posted by another channel with similar cuisine.

However, the viewer may only select a handful of videos from those referrals and decide not to watch them all. In this case, it is also replicating a snowball sampling technique, which is an exponential discriminative snowball sampling. This is a simple example and is open to many direct and indirect factors.

The advantages of snowball sampling are:

• It is cost-effective because the samples are gathered and referred by primary data sources.
• Saves time in finding samples, which can be used to focus on the study.
• The potential new samples can be trusted for reliability as they are obtained with the help of the initial sample population.
• In sensitive studies with crucial topics in which the data is hidden or does not want to contribute can be easily identified.
• Forms a chain of samples, which extends the study’s domain.

The disadvantages of snowball sampling are:

• The study based on snowball sampling may lack the cooperation of new or potential future samples.
• A certain degree of bias is also present with a margin of error.
• It may go against the wish of the potential sample population, sabotaging their agreement and identity.

The key differences between the three are as follows:

• Snowball sampling is where the initial seeds refer to the future or other potential subjects. In contrast, in purposive sampling, the researcher simply selects the sample depending on what is known to them. Still, they choose convenience sampling because of the easy accessibility of the subjects.
• Snowball sampling in research depends on the initial sample for other referrals. In comparison, a researcher uses their knowledge and understanding for purposive sampling, but convenience sampling depends on the availability, access, proximity, geographic location, time and other factors.
• Snowball sampling helps in extending the sample, and purposive sampling helps in selecting the sample size as per the requirement. However, researchers conduct convenience sampling with the easiest relevant sample available.

This article has been a guide to what is Snowball Sampling. We explain its examples, advantages, types, and comparison with purposive & convenience sampling. You may also find some useful articles here -

• Non-Probability Sampling
• Sampling Theory
• Sampling Frame