A researcher trying to understand a specialist workflow may struggle to find the people who perform it. Job titles vary, the relevant community has no public list, and a general recruitment filter produces only approximate matches. One useful conversation can provide a route to the next through an introduction.
Snowball sampling, also called chain-referral sampling, recruits through recommendations from participants or other relevant contacts. It is commonly used to reach people or cases that are difficult to identify through a conventional sampling frame.
Use introductions to widen access
Begin with several suitable contacts where possible, chosen to reach different parts of the population of interest. In a fictional study of museum collection work, starting only with staff at large national institutions could leave small regional museums and independent conservators outside the referral network.
BetterEvaluation’s account of snowball sampling describes using knowledgeable contacts to identify further informative cases. Each recommendation is a lead, not proof of eligibility. Apply the study’s criteria before enrolling someone.
Ask about people whose circumstances or approaches differ, while avoiding pressure to name individuals. Combining referrals with other recruitment routes can help reach cases beyond the original network.
Recognise the sample’s network shape
People often know others with similar roles, organisations or experiences. Well-connected members may be easier to reach, while isolated people remain absent. Several referral chains can therefore produce a sample with considerable internal similarity despite a respectable participant count.
Record enough about recruitment to understand whether one network dominates, while limiting sensitive referral information. Review the developing sample against the experiences the study needs to include and seek different starting points when gaps become apparent.
Ordinary snowball sampling is a non-probability method: inclusion probabilities are generally unknown. It should not be described as representative of a wider population merely because recruitment continued through many waves. More specialised network-sampling approaches involve additional design and analytical assumptions and are not interchangeable with informal referrals.
Protect the choice to participate
Asking a contact to forward an invitation can be preferable to collecting another person’s details without their knowledge. Make it possible for the recipient to decline without explaining themselves to the referrer, especially where the subject or relationship is sensitive.
The method suits exploratory research and studies seeking particular experiences, provided the sample’s boundaries are clear. Purposive sampling can help define which cases are most informative within the recruitment process.
In the report, explain where the chains began, what kinds of participants they reached and which perspectives may remain missing. The introduction helps gain access; the research still needs to earn the interpretation that follows.