What it shows
Snowball sampling recruits through participants’ own networks. Two seeds are recruited first; each refers up to two acquaintances not already in the study, and those recruits do the same, for three waves.
Gray lines are acquaintance ties and arrows show who referred whom. The method can reach people no list would include, and it cannot reach anyone without a tie to those already recruited.
How to read it
The dark units are the two seeds. Each later wave is drawn a little lighter.
Follow the arrows outward from a seed to trace one chain of referrals.
Units with no tie to the sample are never reached, however many waves are run.
The numbers behind it
- Population of 120. Each unit is tied to its right, lower and lower-right neighbor with probability 0.45, set by seeded pseudorandom draws.
- 2 seeds, each recruit referring up to two acquaintances, for three waves.
- Sample: 2 seeds, 4 in wave 1, 7 in wave 2 and 9 in wave 3, 22 in all.
Ways to use it in teaching
Suited to qualitative methods classes on research with groups that have no list to sample from.
Ask how the choice of seeds could shape everyone recruited after them.
Assign Goodman (1961), which formalized snowball sampling, and ask how his design (a random initial sample, each person naming a fixed number of others) differs from the seeds-and-referrals version most qualitative studies use.
For when each method suits a study, and how to justify the choice in a methodology chapter, see our guide to sampling methods.
Sources
Goodman, L. A. (1961), "Snowball sampling", Annals of Mathematical Statistics 32(1): 148–170, doi:10.1214/aoms/1177705148.
