In quantitative research, sampling is often about representativeness: selecting people randomly so that results can be generalised to a population. Qualitative research works differently. Its goal is usually depth, meaning and understanding of experiences, processes or contexts. So instead of asking “Is my sample representative?”, qualitative researchers ask “Does my sample give me rich, relevant information to answer my research question?”
Two of the most widely used approaches in qualitative research are purposive sampling and snowball sampling. Both are legitimate and well established, but both are often described vaguely or used without a clear rationale. Examiners regularly ask students to justify who they interviewed and why. This post explains what these methods are, when to use each one, how to apply them correctly, and how to report them convincingly.
Why qualitative sampling is different
Qualitative studies typically involve smaller samples chosen deliberately. The aim is not statistical generalisation but information richness: participants who can provide detailed insight into the phenomenon you are studying. Qualitative findings may still be transferable to similar contexts, but this depends on rich description of the setting and participants, not on random selection.
Because selection is deliberate, the logic behind your choices must be transparent. A sampling strategy that is clearly linked to your research question is a sign of rigour, not a weakness.
What is purposive sampling?
Purposive sampling (also called purposeful or judgemental sampling) means selecting participants deliberately because they have characteristics, experiences or knowledge relevant to your research question. The researcher decides the selection criteria in advance, based on the study's aims.
For example, if you are studying how first-generation university students adjust to postgraduate study, you would deliberately select postgraduate students who are the first in their family to attend university. Randomly selecting any student would waste time on people who cannot speak to the experience you are studying.
Common types of purposive sampling
Purposive sampling is not one single technique. Methodological writers, especially Michael Quinn Patton, have described many strategies. Some of the most commonly used are:
1. Criterion sampling
Select all participants who meet pre-defined criteria, such as “registered nurses with at least five years of experience in intensive care”. This is probably the most common form in PhD research.
2. Maximum variation sampling
Deliberately select participants who differ on important dimensions, such as age, role, sector or region, to capture a wide range of perspectives. Patterns that appear across very different participants are especially meaningful.
3. Homogeneous sampling
Select participants who are similar in key respects, to explore one group's experience in depth. This is often used in focus groups and in interpretative phenomenological analysis.
4. Typical case sampling
Select participants who represent what is normal or average in a setting, to describe a typical experience.
5. Extreme or deviant case sampling
Select unusual cases, such as exceptionally successful or unsuccessful organisations, to learn from outliers.
6. Critical case sampling
Select a case that is particularly important for testing an idea: “If it happens here, it will probably happen anywhere.”
7. Expert sampling
Select people with specialist knowledge, such as policymakers or senior practitioners, often used in expert interviews or Delphi studies.
8. Theoretical sampling
Used in grounded theory, where each new participant is selected based on emerging concepts from earlier analysis, to develop and refine the theory. Sampling and analysis happen together.
Name the specific type you use, rather than only saying “purposive sampling”. This shows examiners you understand the method.
How to apply purposive sampling correctly
Step 1: Start from your research question
Ask: who has experienced or knows about the phenomenon I am studying? What perspectives do I need to answer my question fully?
Step 2: Define clear inclusion and exclusion criteria
Write specific criteria. For example:
- Inclusion: secondary school mathematics teachers in public schools in [region], with at least two years of experience using online teaching platforms.
- Exclusion: teachers on long-term leave during the study period; teachers who work under the researcher's supervision.
Step 3: Decide on the variation you need
Should the sample include different school types, ages or genders? If so, plan a sampling matrix that shows the combinations you aim to include.
Step 4: Identify recruitment routes
Consider professional associations, organisations, community groups, online forums or gatekeepers. Make sure your recruitment methods are approved by your ethics committee.
Step 5: Screen participants
Use a short screening questionnaire or conversation to confirm that each potential participant meets your criteria.
Step 6: Keep a sampling log
Record who was approached, who agreed, who declined (without identifying details if required), and why each participant was selected. This log supports transparency in your methodology chapter.
What is snowball sampling?
Snowball sampling (also called chain referral sampling) is a recruitment method where initial participants help identify or refer other people who meet the study criteria. Those new participants refer others, and so on, so the sample “snowballs”.
Snowball sampling is especially useful when:
- The population is hard to reach or has no sampling frame, such as informal workers, people in small professional niches or members of hidden communities.
- Trust is important, and people are more willing to participate when invited through someone they know.
- The topic is sensitive, and participants may prefer referrals through their own networks.
Snowball sampling is often combined with purposive sampling: you start with purposively selected “seed” participants who meet your criteria, and then ask them to refer others who also meet those criteria.
How to use snowball sampling correctly
1. Choose diverse starting points
A major risk of snowball sampling is that participants refer people just like themselves, producing a narrow sample from a single network. To reduce this, start with several seed participants from different networks, locations or backgrounds.
2. Keep your criteria central
Referrals must still meet your inclusion criteria. Screen every referred person before including them.
3. Protect privacy and voluntariness
In many ethics frameworks, it is preferable for participants to pass your study information to potential participants, who then contact you if interested, rather than giving you others' contact details without their permission. This protects confidentiality and ensures participation is voluntary. Follow your ethics committee's guidance.
4. Monitor the referral chains
Track which participant referred whom (using codes, not names). If most of your sample comes from one chain, consider recruiting new seeds to broaden perspectives.
5. Be aware of power dynamics
If a senior person refers junior colleagues, those colleagues may feel pressure to take part. Make it clear that participation is voluntary and that referrers will not know who participated.
Purposive vs. snowball sampling: a quick comparison
- Purposive: the researcher selects participants based on defined criteria. Best when the population is accessible and identifiable.
- Snowball: participants help recruit other participants. Best when the population is hidden, hard to reach or trust-based.
- Combined: purposive selection of initial participants, followed by snowball referrals that are screened against the same criteria. This is common in practice.
How many participants do I need?
There is no universal number in qualitative research. Sample size depends on your research question, methodology, the diversity of your population and the depth of each interview. Many researchers use the concept of saturation, the point at which new data add little new insight, to guide decisions. Others refer to “information power”, which suggests that a sample needs fewer participants when the study aim is narrow, participants are highly relevant, and the dialogue is strong.
Plan an approximate range based on similar studies and your methodology, and explain how you decided when to stop recruiting.
An example of a well-written sampling description
Here is an example of how a sampling section might read in a thesis. The details are illustrative:
“Participants were selected using criterion-based purposive sampling combined with snowball referrals. Inclusion criteria were: (a) owner-managers of registered micro-enterprises with fewer than ten employees, (b) operating in the food retail sector in [city], and (c) having adopted at least one digital payment system in the previous two years. Because no public list of such businesses was available, five initial participants were recruited through two local business associations and a community market, chosen to vary by gender, business age and neighbourhood. Each participant was given an information sheet to pass on to other eligible owners, who contacted the researcher directly if interested. All referred individuals were screened against the inclusion criteria. Recruitment continued until interviews stopped producing new codes relevant to the research questions, resulting in 18 participants from four separate referral chains. Participant characteristics are summarised in Table 3.2.”
This description tells the reader exactly who was included, why, how they were found, how diversity was built in, how referrals were managed ethically, and why recruitment stopped. It also anticipates a common examiner concern by mentioning that the sample came from several referral chains rather than one.
Common mistakes to avoid
- Calling convenience sampling “purposive”. If you interviewed whoever was available, without clear criteria, be honest about it.
- Vague criteria, such as “people with experience of the topic”.
- No rationale linking the sampling strategy to the research question.
- Single-network snowball samples that miss important perspectives.
- Ignoring ethics issues in referrals, such as sharing contact details without consent.
- Claiming statistical generalisability from a purposive or snowball sample.
How to report your sampling strategy
In your methodology chapter, include:
- The sampling approach, with the specific type named and a citation.
- Why this approach fits your research question and methodology.
- Inclusion and exclusion criteria.
- Recruitment routes and the number of seed participants, if snowball sampling was used.
- How many people were approached, agreed and participated.
- How you decided on sample size and when to stop recruiting.
- A table describing participant characteristics, using pseudonyms or codes.
- Limitations, such as possible network bias in snowball samples.
Final thoughts
Purposive and snowball sampling are powerful tools for qualitative research when they are used deliberately and transparently. Start from your research question, define clear criteria, choose the type of purposive sampling that fits your aims, and use snowball referrals carefully when the population is hard to reach. Keep a record of every decision. When you can explain exactly who you selected and why, your sample becomes one of the strengths of your study rather than a point of criticism.
Are you recruiting participants for a qualitative study? Share your sampling challenges in the comments.