Snowball Sampling

Snowball sampling uses a small pool of initial informants to nominate other participants who meet the eligibility criteria for a study. The name reflects an analogy to a snowball increasing in size as it rolls downhill.

This approach to locating research participants is almost always used as a form of nonprobability sampling (although some epidemiological research [Page 816]applies techniques from social network analysis to variations on snowball sampling as a way to estimate the total size of populations). Snowball sampling is a useful way to pursue the goals of purposive sampling in many situations where there are no lists or other obvious sources for locating members of the population of interest, but it does require that the participants are likely to know others who share the characteristics that make them eligible for inclusion in the study. This method is particularly useful for locating hidden populations, where there is no way to know the total size of the overall population, such as samples of the homeless or users of illegal drugs.

The typical process for a snowball sample begins with interviewing an initial set of research participants who serve as informants about not only the research topic but also about other potential participants. In some cases, the process of snowballing that follows the initial interviews is indirect in the sense that these original sources mostly supply information about how to locate others like themselves; that is, where such people are likely to congregate, how to recognize them, and so on. In classic snowball sampling, however, the initial informant often assists in recruiting additional participants into the study. Depending on the number of people sought, this process of using earlier informants to locate new informants may go through several rounds. For example, a single initial informant might put the researcher in touch with three other sources who might assist in locating seven more new sources, and so on.

In practice, snowball sampling poses a distinct risk of capturing a biased subset of the total population of potential participants because any eligible participants who are not linked to the original set of informants will not be accessible for inclusion in the study. The best defense against this problem is to begin with a set of initial informants that are as diverse as possible. This variation on maximum diversity sampling increases the likelihood that the subsequent links in the snowballing process will reach different segments of the total set of eligible participants.