What is Snowball Sampling?
Snowball sampling is a non‑probability recruitment method where existing study participants recruit future participants from their networks. It’s commonly used to reach hard‑to‑find or tightly knit groups when a sampling frame is unavailable.
In snowball sampling you begin with a small number of initial participants (seeds). After they complete the survey, you ask them to refer people they know who meet the study criteria. Those referrals recruit further respondents, and the sample grows like a rolling snowball. The technique is quick and inexpensive and works well for populations linked by social networks (e.g., recent migrants, service users, or community groups). However, because people recruit within their own networks the sample is not randomly selected and can be biased by who the seeds know, network structure and patterns of social similarity (homophily). A more structured variant, respondent‑driven sampling (RDS), adds recruitment limits and statistical weighting to help correct some biases.
Usage example
A local council wants feedback from newly arrived refugee families about local services. They start with three families connected to a community centre (the seeds) and ask each family to invite two other families to take the survey. Responses expand through referrals until the council has enough responses to understand common concerns across that community.
Practical application
Snowball sampling matters because it enables organisations to gather input from communities that are difficult to reach through standard random sampling or public invitations. For Hearo users, it can rapidly increase participation among multilingual or tight‑knit groups by leveraging trusted personal or community referrals. Use it when inclusion and reach are more important than producing statistically representative estimates. To make results more reliable: select diverse seeds, track referral chains and referral sources, limit the number of referrals per participant, and be transparent about limitations when reporting findings. Combining snowball sampling with other outreach (community partner invitations, targeted outreach in different languages) reduces bias and improves coverage.
FAQ
Is snowball sampling representative of the whole population?
No. Snowball sampling is a non‑probability method and does not guarantee a representative sample. Because respondents recruit from their own networks, some groups can be over‑ or under‑represented. Use it to gain insights and reach hidden populations, but avoid treating findings as statistically generalisable without additional methods or weighting.
How can I reduce bias when using snowball sampling?
Start with multiple, diverse seeds from different segments of the community; limit how many people each participant can refer; track who recruited whom and how respondents heard about the survey; and supplement referrals with other outreach channels in the relevant languages. Where possible, consider RDS or apply post‑hoc weighting if you have population benchmarks.
When is snowball sampling a good choice?
Choose snowball sampling when you need to reach hard‑to‑contact or networked populations (e.g., recent migrants, informal workers, niche community groups), when there’s no reliable sampling frame, or when rapid, low‑cost recruitment is needed. It’s less suitable when you need precise, generalisable prevalence estimates.
How can Hearo support a snowball sampling approach?
Use a single multilingual survey link that participants can share with contacts in their preferred language. Add an optional question asking where they heard about the survey or who referred them to capture referral chains. Provide clear instructions in each language so referrers know whom to invite, and monitor response patterns by language and referral source to check for coverage gaps.