What is snowball sampling?

Contacting the right people to participate in your research is vital – but finding these subjects can be tricky in some cases.

People may be unwilling to communicate or challenging to find for several reasons. Snowball sampling is a recruitment technique designed for these specific situations since it uses a non-probability sampling method.

Instead of including a general group of potential participants, snowball sampling involves current participants helping to recruit other potential subjects.

The name comes from the way participants gather new people, increasing the size of the group. In this sense, it’s similar to the way a rolling snowball grows in size. It’s also sometimes known as chain-referral sampling.

There are three types of snowball sampling. They vary slightly, but they all rely on referrals from participants.

Linear snowball sampling

This method forms a sample group with a single subject. They provide contact information for one other person who would be appropriate for the study. That person is invited to join and provide one referral themselves. This continues until enough people have been recruited for the study.

Exponential non-discriminative snowball sampling

As with linear snowball sampling, this method starts with a single participant. They provide contact information for multiple potential subjects who are invited to participate. These people are then asked to provide referrals to multiple other potential participants.

This pattern continues until enough participants have joined the study. Naturally, it’s the fastest method for filling a study’s requirements.

Exponential discriminative snowball sampling

This approach is a combination of the previous two. Each participant provides multiple referrals to people who may be interested in joining the study. Out of that group, one person is invited to participate based on the study’s unique requirements. This enables researchers to narrow down the participants based on the characteristics they’re looking for.

This type of snowball sampling takes the most time, but it allows researchers to make the most detailed choices about who participates in the study.

Research conducted with the snowball sampling method cannot be used in quantitative research. However, it can be very helpful in qualitative research (when applied carefully) to provide insight into certain communities or situations.

Each of the three snowball sampling methods includes one crucial common factor: study participants refer others for the study. This can be an advantageous way of making contact with members of a community when it might otherwise be difficult.

The primary reason researchers use snowball sampling is to gain access to populations that are difficult to contact. This can be for many different reasons.

One example is people who are members of an “elite” group. This may include very wealthy individuals or members of an exclusive club. Such people may be reluctant to reveal their contact information to the general public, approach their peers, or even be open about their status.

Another group that may be difficult to contact is people who are secretive about their identity or condition.

Examples of this might be homosexual clergy members or people who have an infection that carries a social stigma, such as HIV. Being referred to the study by other members of the same group may make them more likely to feel open to speaking with researchers.

Snowball sampling can also be used to contact people who are difficult to find because of their living situation, such as refugees or homeless people. In many cases, there are no public records of these people.

It can be very useful when someone from within this community reaches out to them and asks about their willingness to participate in research.

This approach is used when researchers have trouble locating certain groups. This can occur for a variety of reasons, such as social stigma, a lack of public information about the group, or a desire for people to protect their identity.

Snowball sampling can’t be used for statistical studies because it won’t deliver a representative sample. However, it can be very helpful in qualitative research.

Here’s an example of how this method can be used.

If a researcher wants to speak to people who have rented housing in a certain area for the past five years, they may find it difficult to locate enough people who meet the study’s requirements. Rental information is not a public record.

People are often familiar with their friends’ living situations, so they’re likely to be able to provide names of people in the same situation as them.

If researchers can identify just a few individuals who meet the guidelines, they can ask them to provide referrals for other people in similar circumstances. The researchers can then continue this process until they have found enough people for their research.

This approach lends itself to certain types of research that focus on communities that may be difficult to locate.

For example, researchers conducting medical studies of rare diseases often struggle to find participants.

Identifying people with specific rare conditions can be difficult, especially given patients’ legal and ethical rights to privacy. However, other people with a rare disease are likely to have contact with fellow sufferers and may be motivated to help researchers in their work.

Social research is another area of study that frequently uses snowball sampling. It can be difficult to recruit enough people to provide a robust population sample. This is especially true if the group is stigmatized or disadvantaged in some way as they are more likely to be reluctant to talk to researchers.

Snowball sampling provides a way into these protective communities.

Examples of snowball sampling

The specific situations that call for snowball sampling demonstrate frustrating barriers to other data collection methods. Here are some examples:

• Populations without an official record of members. Some demographics don’t list who they are and how to reach them. For example, demographic information about homeless people is difficult to locate.

• People who don’t want to be identified. Sometimes, researchers might want to study people who are hesitant to reveal information about themselves. Examples include people who are involved in criminalized behavior, such as sex work, people who have suffered a traumatic event, such as sexual assault, or people who face judgment in their community. 

• People who are secretive about their personal identity. This could include people involved in a cult or other extreme, insular community. They may mistrust outsiders and be reluctant to expose their fellow members to scrutiny.

Snowball sampling offers advantages to researchers. Here are some of the reasons they might choose this method:

• Provides access to hidden groups. People who are motivated to conceal their identities are naturally difficult to locate by people outside their community. Referrals from trusted individuals might open doors.

• Quicker and less expensive. Working from referrals to contact potential participants takes less time, effort, and funding than other approaches.

• More flexible approach. Instead of rejecting potential participants based on pre-set rules and random selection, snowball sampling enables researchers to follow the recommendations of community members.

No method is without its disadvantages. Here are some to consider when using snowball sampling.

• Not all members of a group have an equal chance of being selected. The sample group is not chosen through random selection, making it susceptible to a margin of error. Researchers can’t assume they are studying an equal cross-section of the group, preventing them from making statistical inferences.

• Possible lack of cooperation. There is no guarantee that potential participants will be interested in speaking to or cooperating with researchers.

• Potential for sample bias. Because participants tend to refer people they know, researchers run the risk of sample bias. This is where the group includes people of similar backgrounds and circumstances instead of a random representation of the group.

Does snowball sampling reduce bias?

There is an increased risk of sample bias and margin of error with snowball sampling. This method doesn’t use random selection, and the participants are likely to refer people who are similar to themselves. For this reason, the results may not fully represent the population.

Snowball sampling is an important tool that researchers can use when they are developing a study. Although it has drawbacks and limitations, it can also enable you to reach people who would otherwise be difficult, if not impossible, to find.