Course contentsShow
← Data Science
Lesson 241 of 2,1457. Sampling and EstimationFree lesson

Snowball Sampling

Use existing subjects to recruit additional participants from their networks.

Snowball Sampling

What you'll learn: How to grow your sample by having participants recruit others from their own networks.

What Is Snowball Sampling?

Snowball sampling is a non-probability sampling method where you start with a small group of participants and ask each one to refer or recruit additional people they know who meet your study criteria. Like a snowball rolling down a hill that picks up more snow, your sample grows by accumulating new participants through referrals.

How It Works

You begin with a few "seed" participants who fit your target population. After collecting data from them, you ask: "Do you know anyone else who might qualify for this study?" Those referrals become your next wave of participants, who in turn refer others, and so on.

This creates chains of referrals that expand your sample organically through existing social networks.

When to Use Snowball Sampling

This method shines when studying hard-to-reach or hidden populations:

  • People with rare conditions or experiences
  • Members of tight-knit communities
  • Populations without public lists (e.g., undocumented workers, people in recovery)
  • Groups where trust and personal connections matter

Key Limitations

Because participants recruit from their own networks, your sample will likely not be representative of the broader population. People tend to know others similar to themselves, creating homogeneous clusters and potential bias. You cannot generalize findings with statistical confidence.

Practical Example

Imagine studying rare disease patients. You find three patients initially, interview them, and each refers two friends with the same condition. Now you have six more. Those six refer others, and your sample expands through personal connections where formal databases don't exist.

Key Takeaway: Snowball sampling uses referrals from existing participants to reach hidden or hard-to-access populations, but produces non-representative samples that limit generalization.