Sometimes, researchers find it difficult to access the relevant variables for their study and build a sample. When this happens, you can explore snowballing sampling instead of stalling your systematic investigation or ending it abruptly. So how does snowball sampling work?
For a better understanding, you need to cast your mind to winter—remember how a rolling snowball gathers the snow on its path? This is exactly how snowball sampling works. In other words, the researcher depends on a small number of participants to help him identify other potential research subjects.
This article discusses the different types of snowball sampling, plus common use cases for this non-probability sampling method.
Also known as chain-referral sampling, snowball sampling is a non-probability technique that allows the researcher to discover variables with rare characteristics. Here, the researcher finds a handful of suitable primary data sources and uses them to discover similar variables for the study—think of it as a referral technique that thrives on the principle of like-minded attraction.
For instance, let’s say you want to collect responses from patients who suffer from a rare type of cancer. In this case, other sampling techniques might prove inadequate for gathering relevant subjects—you cannot just walk into the hospital and request patients’ contact information or medical records. What you can do, however, is put out a call to speak with one or two patients with the condition, and then ask them to refer you to other potential subjects who might be willing to participate in your study. Patients can continue a chain-referral process until there’s enough population to form your sample frame.
Read: Convenience Sampling: Definition, Applications, Examples
Snowball sampling gets its name from a distinct feature of snowballs during the winter.
One of the most exciting parts of winter is the snow, and the different games you play with snowballs. You’d notice that when you form a tiny snowball and set it rolling, it continues to pick up more snow along its path until it forms a large ball of snow, ready for the games. This is what happens with snowball sampling.
First, the researcher forms a small snowball by identifying a handful of potential participants for the systematic investigation. Then, s/he set the ball rolling by asking existing participants to “pick up” similar variables. As they continue to refer new subjects, the sample population or snowball grows bigger until you have enough data to work with.
Read: What is Stratified Sampling? Advantages, Examples, Definition, Types
In this section, we’d discuss three common types of snowball sampling which are;
In linear snowball sampling, the researcher depends on a straight-line referral sequence that starts with a single variable. This variable provides information about one other potential research subject, and it goes on until the desired sample population is achieved.
The linear snowball sampling model depends on one referral per subject. Here’s an image that illustrates this:
Exponential non-discriminative snowball sampling runs on a geometric chain sampling sequence. Here, the formation of a sample population starts with finding one suitable participant. Now, this participant goes ahead to refer multiple potential research subjects, and the chain continues until you have enough participants for your study.
This is quite similar to exponential non-discriminative snowball sampling. However, in this case, the researcher screens potential variables before accepting them into the sample population. Only variables that meet the screening criteria are allowed to participate in the systematic investigation.
Snowball sampling is best used in contexts with a specific and relatively small population that is difficult to identify or locate. Because of its non-probability nature, it is one of the best data collection techniques for exploratory or qualitative research.
There are two important steps you should take when it comes to snowball sampling. First, you need to identify potential subjects that closely fit your research, and can help you achieve your aims and objectives. Typically, you’d find only one or two possible variables in the first instance.
Next, ask these initial variables to recruit other potential participants who share the same characteristics as them. To get the most response, you can throw in incentives that encourage participants to refer others.
1. Accelerated Sampling: With the snowball technique, you can easily find suitable participants for your systematic investigation. By depending on the chain-referral system, you can discover variables that share rare traits that are relevant to your research process.
2. Cost-effective Sampling: Since you’re not scouting for research participants by yourself, you don’t have to invest lots of money in the sampling process. You only depend on the referrals obtained from your primary data sources.
3. Snowball sampling also allows you to collect responses from people who would have hesitated to take part in your research. Since there’s an existing relationship between each variable and their referral, the referrals are quite eager to participate in the data collection process. For example, someone with a rare medical condition would feel more comfortable speaking with a patient with a similar challenge.
4. This sampling technique requires little planning and fewer workforce compared to other sampling methods.
5. Snowball sampling can help you to discover other rare characteristics about the variables in your sample population.
1. With snowball sampling, it is challenging to identify any sampling errors or make inferences about the sampling population.
2. Snowball sampling removes the researcher from the center of the sampling process. This means that the researcher has little or no control over the sampling method, and relies mainly on the referrals from already-identified participants.
3. It can trigger sampling bias in your study due to a lack of representation of the population of interest.
4. Snowball sampling can increase the margin of error of your research results, leading to more variations between the sample results and the population of interest.
The common line that runs through all applications of snowball sampling is the difficulty in locating suitable variables for the research. This difficulty prompts the researcher to create some sort of data collection bypass for the systematic investigation by depending on referrals for the sample population.
Common reasons why you might resort to snowball sampling include:
Identifying the right sample size is important because, if it’s too little, it can lead to inaccurate results. However, if it’s too large, you might end up wasting valuable time and resources.
For snowball sampling, the sample size is usually heavily dependent on whether the population is known or not.
1. If the population is finite or known you can use online calculators like calculator.net to find the exact sample size.
2. If the population is infinite or unknown (i.e you have no idea what the population could be), you can use the Cochran Formula. Find out more here
1. Medical researchers use snowball sampling to collect responses from patients with rare diseases. Suppose you’re researching the symptoms of a condition like a porphyria. People with this disease might be unwilling to speak with a researcher about their illness. However, with snowball sampling, you can use the chain-referral system to draw them out and gather the data you need.
2. Let’s say you’re researching heroin usage in a particular location. On your own, it might be difficult to draw out people who use this substance for several reasons. For example, they might believe you’re part of law enforcement, and speaking with you would get them arrested. However, if you can contact a small part of your target population, they could help you contact possible participants for your study.
3. If your study requires responses from ex-convicts, it would be difficult to get an adequate number of people who are willing to participate in your study. However, if you could invite a few ex-convicts to be part of your research, they could help you get in touch with others in their group.
4. Suppose you’re conducting a study on the homeless in society. It may be difficult to obtain a list of homeless individuals and their contact information. However, if you can get in touch with a handful of homeless individuals, they could help refer other homeless people for your study.
Researchers adopt snowball sampling when it is difficult to gather the best-fit variables to form the sample population of the study. This happens for a number of reasons including difficulty in obtaining the contact information of subgroups in the target population, social stigma, and confidentiality clauses that prevent people in the group from providing any information.
The primary difference between purposive sampling and snowball sampling is how the researcher gathers participants for the study.
In purposive sampling, the researcher uses their discretion to select suitable participants for the study, based on their knowledge of the context of the systematic investigation. However, in snowball sampling, the researcher depends on existing research participants to help identify other potential subjects.
There is no specific formula for snowball sampling in research. If you adopt the linear sampling method, then each of the existing variables would have to provide one possible research participant. For the exponential methods, existing variables can identify as many possible participants as they can find.
Many business leaders adopt snowball sampling for high-level research because it helps them get responses from a specialized audience. Outcomes from snowball sampling are highly relevant to the research context and can be easily extrapolated to a wider audience.
It’s best to use this sampling method when you’re dealing with a small population. In the case of a larger audience, opt for probability sampling methods that guarantee the representation of different subgroups in your sample.
You may also like:
In this post, we will discuss extensively what acceptance sampling is and when it is applied.
In this article, you’ll discover what purposive sampling means and how to implement this research method in your systematic study
In this article, we are going to discuss the concept of non-probability sampling, its advantages and disadvantages, and where it can be used
In this article, we will discuss what population of interest means, how it differs from a parameter of interest, how to determine the...