The Benefits Of Snowball Sampling [+ Examples]
Written by: Phil Hesketh
Most UX research studies can be effectively carried out using simple recruiting strategies which will yield enough consenting participants to meet the criteria for a study.
But sometimes research requires the insights of a small and specific subset of the population, who are often challenging to find by traditional means.
Snowball sampling isn’t a term you hear thrown around much in large-scale enterprise research, but if you’re looking for a group of people with hard-to-find characteristics, this method of recruiting could be just what you’ve been looking for.
In this article we’ll take a look at how to ethically recruit using the snowball recruiting method, and some of the pros and cons around using the feedback you collect to inform your research.
What is the snowball recruitment method?
The snowball sampling recruitment strategy (aka network sampling, chain sampling, referral sampling, or chain referral sampling) is a non-probability sampling method for participant recruitment that taps into existing participants’ social networks to find more people who are like them.
It’s typically used when researchers need to talk to people who are in niche social groups, who have very defined traits, or are otherwise “hidden” and difficult to reach out to using more traditional methods of recruitment.
People with specific criminal records
People with rare diseases
People with hidden sexual orientations that could potentially damage their career
The primary purpose behind snowball sampling is to use social trust when identifying people in a community that otherwise would be tough to find. Most often, it’s helpful when you’re asking unique qualitative questions to a population that (for whatever reason) doesn’t self-identify easily.
Researchers are always conscious of the trust we’re building (or lacking) with a community. That trust is crucial, but it takes time, energy, and effort to build. Understanding the feasibility and the amount of effort we can expend goes a long way toward matching the recruitment method with the project goals.
The first question is one of efficacy. Are we certain that members of this particular group will give us insight we’re not able to get elsewhere?
If the answer is yes, we can think about how to make contact and move into a successful position. Suppose we’re interested in researching a population with whom we have low trust (for example, a politically persecuted group). In that case, we’ll be more successful if we communicate in ways that establish trust, such as speaking one-on-one (or in a small group in some cases), meeting face-to-face, active listening, gathering context, and focusing on their interests, not only what we hope to gain.
The second question is one of potential. Do we have the resources, the time, and the energy to start this process well enough that the snowball can continue to grow? We will spend 90% of our effort getting the snowball started, and if we establish a trusting foundation, we will set ourselves up for future success.
Third, we’re trained to always be ensuring beneficence to the populations we’re working with. What’s in it for them? Describing how more insight or attention will connect to some reward will go a long way towards helping folks identify others in their network. Those benefits could look like more advocacy for their community, addressing misconceptions, or advancing an idea or cause. Of course, ensuring that these promises are feasible when made and followed through on is essential.
So, the third question is one of mutuality. Does this study clearly benefit the population we’re trying to discover?
I focused my thoughts ethically because effective human-centered research is a byproduct of strong relationships. Strong relationships come from established goodwill. And goodwill is built when we look for opportunities to build trust early and often.
—Jason Garrison, Senior UX Researcher
In theory, all you need to do is find one participant for your research study, and ask if they know anyone else who fits the same required profile for the research.
But in practice, researchers using this method also need to be highly skilled at screening, building trust with uneasy or at-risk people, and know how to approach sensitive topics with care, responsibility, and understanding.
Examples of snowball recruiting
Researchers might choose this method of recruiting when they’re faced with tricky sampling situations such as needing to make contact with:
People with rare diseases who are difficult to locate
People with no public record, such as illegal immigrants or the homeless
Members of elite or secret clubs
Trauma or assault victims
Religious or political extremists or activists
People with occupations that might be illegal, such as hackers, graffiti artists, or sex workers
Snowball sampling is a recruitment strategy that can be useful for researchers who are trying to understand the reasons why people join a secretive religious cult, and what it’s like for members on the inside.
But it’s extremely difficult to collect any data or information unless you can find one cult member that is willing to talk to you, and then refer you on to other members who might be willing to share their experiences.
People that carry a weighty social stigma, such as homosexual clergy members who may have contracted HIV.
For obvious religious and social standing reasons, this segment is both difficult to find and difficult for researchers to speak with directly. So being referred more study participants by an existing participant might be the only way to increase your sample size.
Researchers might be trying to study the growth of the homeless population in London, but this segment of society moves around and has no fixed abode.
There is no list of the details of every homeless person in London, but if researchers can speak to one or two homeless individuals, the snowball recruitment method can help them get a better picture of this demographic.
Ethics and safety concerns
It’s important to note that with many “hidden” populations, such as the examples mentioned above, there needs to be a LOT of safeguarding carried out before any research or recruitment takes place — for both the researchers’ and the participants’ sakes.
It’s best to contact these audiences through a third party (such as a charity or non-profit / school or community center etc.) so that there is a support network in place for them when you leave as a researcher.
These participants could potentially be reliving very triggering experiences, or the information they share with you could have severe consequences for them which could present a significant risk for them, and could also put the researcher in harm's way.
Different types of snowball recruitment
These are the main snowball recruiting methods for researchers. Although their approach is slightly different, they all have the referral factor in common. You can use whichever one makes the most sense for your particular research study.
Exponential non discriminative snowball sampling
This is where an initial participant is recruited, and they provide multiple referrals to a researcher. Each new referral then provides multiple new referrals until researchers have the sample size they need to effectively carry out their study.
Having participants refer multiple people from their network is a fast-track for researchers to increase their participant pool. On the downside, researchers will need to carefully screen each new referral that comes in to check that they meet the specific criteria.
Exponential discriminative snowball sampling
This snowball sampling framework is similar to the above recruitment strategy, with each participant giving researchers multiple referrals. But with the discriminative method, researchers will only recruit one new referral out of each new group of people that is referred to them by a current participant.
This process of screening and narrowing down referrals takes time, but it also ensures that researchers can make better choices about inviting participants who are relevant to the study.
Linear snowball sampling
With the linear sampling method, an initial participant will refer just one other person to researchers. If this person is invited to the study, they in turn will be invited to refer to another person from their network, forming a linear “chain” of participants.
This is a much slower method than the exponential recruiting strategies, and it can take a while to find the amount of people to potentially participate for each study — but these people can also turn out to be a more accurate fit for a research study.
A blend of field sample and custom estimation procedures, the respondent-driven recruiting strategy is a relatively new approach to snowball recruiting.
It helps researchers avoid some of the biases that can occur with snowball sampling - for example, initial participants protecting their friends by not referring them on to researchers.
With this method, recruited participants pass on information about other participants to researchers, instead of referring the person directly. This can protect at-risk individuals, and also give researchers valuable insights for their study.
Virtual snowball sampling
This type of sampling relies on virtual networks of participants from platforms such as Facebook.
The advantages for researchers is that it can be faster than other types of snowball sampling, and there is a higher response rate from “hard to find” participants with this method, especially if they wish to remain anonymous.
Peer Esteem Snowballing (PEST)
This is another version of snowball sampling used to gather information from small populations of experts. It has a few advantages over the other methods.
PEST can help researchers:
Improve participant response rates
Reduce selection bias
Take network boundaries into account
Create an estimate of population size
Allow for clustering of expert opinions on the basis of their referral network
How do you recruit participants using snowball sampling?
The type of research, together with your timeframe and budget, will help point you towards the right snowball sampling method for each study.
Identify initial participants
First you’ll need to identify and qualify an initial participant who will become the starting point for your referral chain.
Contact these participants
Once you’ve found your first participant, reach out to them to ask if they would be interested in helping you conduct research. This can be done via email or social media, or in person depending on the social profile you’re studying.
Tell them briefly about your goals for the study, and reassure them about things like anonymity, data safety, and the fact that providing informed consent for the study is optional and able to be revoked at any time.
Encourage them to refer you to similar participants
Once you’ve built trust with your initial participant, you can encourage them to refer you to another person (or people) within their network that fit the same profile.
Having a person refer somebody they personally know to you can increase response rates, improve accuracy for your participant pool, and help to quickly snowball your sample numbers.
Evaluate referrals for suitability
Whether you’re using a linear or discriminative sampling variation, you’ll need to screen and approve all of the referrals that come your way. This is especially important if you’re going with the exponential discriminatory sampling strategy.
Check and analyze responses from your referral as they come in. This can help you pinpoint whether you’ve been speaking to the right people, or whether you need to continue your outreach.
Establish a margin of error
All of the snowball sampling methods can create biases and sampling errors in your data. You should calculate a margin of error to ensure that your findings are as accurate as possible.
Repeat the above steps
Snowball sampling involves relatively simple steps to get things rolling (pun intended), so if you’ve tapped into the correct social network — all you’ll need to do is keep the referrals coming in, evaluate research participants and their responses, and repeat.
Pros and cons of snowball recruiting
This type of sampling is useful for qualitative research to uncover deeper insights into hidden populations, but it has its limitations, including multiple biases and the reluctance of ideal participants to be recruited for studies.
Advantages of snowball sampling
Cost-effective for researchers on a budget
Saves time and resources trying to figure out how to populate your sample pool
Faster to find samples from small and specific social populations
Reduces the risk of spending time speaking to non-relevant participants
Can help expand the geographical scope of a study
Can increase the number of responses
Can improve quality of responses
Can be carried out in-person or remotely
Disadvantages of snowball sampling
Can be biased in respect of sample factors like age, gender, socioeconomic stature
Ideal participants might not have access to email, internet, or phone
Initial participants must be carefully screened, as they will have a major impact on the quality of referrals
Referrals might refuse to be contacted
With an unknown total population size, there may not be enough ideal participants for researchers to achieve their target participant number in a given study
Participants might not actually give an accurate picture of the target population
Potential ethical issues might prevent researchers speaking directly with ideal referrals
Possible doubts about the quality of informed consent (e.g. from substance abuse participants), and in particular, their "decision-making capacity"
Participants will typically know each other, which means potential participants who are isolated can not be reached
Researchers have little control over the sampling process
Snowball sampling is a handy recruitment strategy for every UX researcher’s toolkit. Although it has its pros, cons, and limitations, it can help researchers more easily speak with specific social groups who would otherwise be impossible to find.
If you’re looking to carry out UX research studies for your organization, Consent Kit can help you formalize your recruitment process, organize and manage participant data with ease, and ensure every study is accessible, compliant, and efficient. Try it free for 14 days.