What is Snowball Sampling?
Snowball sampling is a non-probability method for acquiring a sample that uses participants to recruit additional participants. Researchers call it snowball sampling because if the initial participant recruits two more, and those two recruits each bring in two more, and so on, the number of participants can grow exponentially like a rolling snowball. This method is also known as chain sampling or network sampling.
Use the snowball sampling method when working with a population that is difficult to find. For example, use it for projects that work with criminals, drug addicts, undocumented workers, have a severe social stigma, involves a sensitive topic, or a rare condition. Other uses are for tiny populations and those that are highly dispersed.
In general, this method is a suitable when researchers will have a hard time finding these people themselves, but population members are likely to know others. In some cases, current participants not only identify additional subjects but can also assure them of the study’s importance and calm their fears about confidentiality.
Learn more about Types of Sampling Methods in Research.
Snowball Sampling Example
To understand the potential benefits of this method, the following is a snowball sampling example from an actual study. (Kaplan, Korf, & Sterk, 1987)
In this study, researchers effectively used snowball sampling to understand heroin users in two Dutch cities. This population is hard to find because their activities are both sensitive and illegal. Consequently, researchers didn’t know much about their social contexts and lifestyles.
The study used this method to build up a sample despite this population’s elusiveness. They discovered two distinct subpopulations of heroin users with differing lifestyles, contexts, and life cycles. They also explored the transition from adolescent peer groups to criminality, mapped out the nature of the social network, and measured interaction frequencies.
This type of information would be virtually impossible to obtain without this example study using the snowball sampling method.
Types of Snowball Sampling
The snowball sampling method comes in three flavors. The differences involve how many subjects each participant finds and how many of them the study includes.
With linear snowball sampling, the first subject finds only one recruit. In turn, that subject finds one more, and so on. Researchers repeat this one-at-a-time process until they obtain their target number.
With the exponential non-discriminative snowball sampling method, the first participant recruits multiple subjects. Each of these new subjects recruits several more subjects. This approach allows the sample to grow exponentially until it reaches the target size.
The exponential discriminative snowball sampling method is like the non-discriminative form because each participant finds several more potential recruits. However, the researchers pick only one of each subject’s referrals, giving the researchers more control over who participates. They make their selections using the study’s goals and objectives as guides.
Advantages and Disadvantages of the Snowball Sampling Method
Snowball sampling has various advantages and disadvantages compared to probability methods, such as random sampling.
Snowball sampling allows researchers to access hard-to-find populations. The exponential forms of the method enable the sample size to expand rapidly. Because it is a non-probability sampling method, researchers don’t need a complete list of the population. The only participation requirement is that the subjects are willing to participate and help you find others.
Consequently, this method allows researchers to obtain a large group of participants quickly and easily, even for a population whose members usually don’t want to be found.
Snowball sampling is a non-probability method. Ultimately, convenience is a factor for this type of method. In this case, the ongoing recruitment process depends on who the current participants already know. Consequently, the sample you obtain is likely to be biased and not represent the population, preventing generalizations to the population. Learn more about Representative Samples and Sampling Bias.
Compared to quota sampling, another non-probability sampling method, the researchers have less control over who they select. However, given that the population is hard-to-find, this tradeoff might be acceptable.
Kaplan, Charles D. Ph.D.; Korf, Dirk Ph.C.; Sterk, Claire Ph.C., Temporal and Social Contexts of Heroin-Using Populations And Illustration of the Snowball Sampling Technique. The Journal of Nervous and Mental Disease: September 1987 – Volume 175 – Issue 9 – p 566-574.