03 · What You Need to Know
How Qualitative Sampling Follows the Question, Methodology, and Emerging Analysis
Qualitative Sampling Begins With the Phenomenon, Not a Magic Number
Before asking how many participants you need, ask what kind of participants or cases can actually provide the evidence required by your question.
A study of first-generation university students' transition experiences needs participants who meet the relevant definition of first-generation students and have experienced the transition being investigated. A study of how academic leaders respond to institutional cyberattacks needs people with direct involvement in such events. Randomly selecting people from an entire university workforce would be inefficient if most have no experience of the phenomenon.
Methodological guidance for qualitative research emphasizes that sampling strategies should produce rich information and remain consistent with the qualitative approach being used. Participant selection is therefore part of the logic of inquiry rather than merely a recruitment procedure.
Purposive Sampling Is Common, but It Is Not One Single Technique
Purposive sampling deliberately selects participants, cases, or other information sources because particular characteristics make them relevant to the study.
This broad principle can be implemented in different ways. A researcher may seek participants who closely fit a specific phenomenon, intentionally include diverse cases, select unusual or especially informative cases, recruit contrasting cases, or modify subsequent sampling in response to emerging analysis.
The important point is that selection has a reason beyond accessibility.
This distinguishes purposive sampling from convenience sampling. A participant can be both highly relevant and conveniently accessible, of course, but the sampling rationale should identify which logic actually drove selection.
Criterion-Based Sampling Selects Participants Who Meet Defined Conditions
Some qualitative studies require all participants to satisfy particular experiential or case-related criteria.
Suppose the research question concerns how faculty members respond after discovering undisclosed generative AI use in student submissions. It may be reasonable to require participants to have personally encountered and handled at least one such case.
This creates a sample directly tied to the phenomenon.
The criteria should still be justified carefully. Adding unrelated requirements such as age, academic rank, employment status, or years of experience without a methodological reason can unnecessarily narrow the available perspectives.
The broader principles for establishing inclusion and exclusion criteria therefore remain relevant to qualitative studies.
Maximum Variation Sampling Seeks Meaningful Diversity
Sometimes the objective is not to make participants as similar as possible but to examine how a phenomenon appears across meaningfully different circumstances.
A researcher studying faculty experiences with AI-assisted assessment might deliberately include instructors from different disciplines, career stages, institutional contexts, or levels of AI experience if those dimensions are expected to illuminate important variation.
This is often described as maximum variation or maximum variation purposive sampling.
The purpose is not to reproduce the demographic proportions of a population. It is to capture analytically meaningful diversity and examine patterns across heterogeneous cases.
Statistical representativeness
Concerns how a sample supports inference to a specified population, commonly through a probability sampling framework.
Purposeful variation
Deliberately includes relevant differences among cases so the qualitative analysis can examine how a phenomenon varies across them.
Homogeneous Sampling Can Be Appropriate When the Question Is Narrow
Not every qualitative study needs maximum variation.
If the objective is to understand a highly specific experience within a clearly bounded group, a relatively homogeneous sample may allow researchers to investigate that experience in greater depth without introducing variation irrelevant to the question.
For example, a phenomenological study might focus specifically on first-year intensive-care nurses who experienced their first patient death during their initial year of practice. The narrow eligibility is defensible if that particular lived experience is the phenomenon under investigation.
The question determines whether similarity or variation is analytically useful.
Sampling Can Be Iterative Rather Than Completely Fixed at the Beginning
Qualitative research often involves interaction between sampling, data collection, and analysis.
Early interviews may reveal a perspective that is poorly represented. Researchers may then seek participants who can illuminate that issue, provide a contrasting case, or help clarify an emerging interpretation.
Methodological guidance describes qualitative data-collection plans as capable of remaining flexible during the study, with subsequent sampling informed by what the developing analysis still needs.
This does not mean changing the sample arbitrarily whenever an interesting participant appears. Iterative decisions should remain connected to the research question, methodology, ethics approval or protocol requirements, and emerging analytical needs.
Theoretical Sampling Has a Specific Methodological Meaning
Theoretical sampling is particularly associated with grounded theory. It should not be used as a generic synonym for purposive sampling.
In theoretical sampling, emerging concepts and categories guide decisions about what data or cases should be collected next to develop, refine, compare, and relate those categories as theory develops.
A researcher might begin interviewing one group, identify an emerging category that needs comparison, and subsequently seek participants or situations capable of testing or elaborating that category.
The sample therefore develops alongside the analysis. Simply deciding beforehand to interview “different types of people” does not by itself constitute theoretical sampling.
Snowball Sampling Can Help Locate Participants Who Are Difficult to Find
Some relevant participants cannot easily be identified through public lists, institutional records, or conventional recruitment channels.
In snowball or chain-referral sampling, initial participants or contacts help identify additional potentially eligible people. This may be useful for rare experiences, sensitive phenomena, dispersed professional groups, or populations whose membership is not easily enumerated.
The method also has consequences. Referral chains reflect social networks. Participants may refer people similar to themselves, highly connected individuals may be easier to reach, and isolated perspectives may remain invisible.
Snowball sampling should therefore be chosen because network-based recruitment solves a genuine access problem, not simply because asking interviewees for names is convenient.
Sampling Strategy Should Match the Qualitative Methodology
“Qualitative research” encompasses multiple methodological traditions with different analytical objectives.
| Research purpose or approach |
Sampling consideration |
Possible question |
| Phenomenological inquiry |
Participants should have direct experience of the phenomenon being investigated |
Who can provide rich first-person accounts of this lived experience? |
| Grounded theory |
Sampling may become increasingly theoretical as emerging categories require comparison and development |
What case or situation should be sampled next to elaborate the developing analysis? |
| Case study |
Case boundaries and the logic for selecting the case or cases require explicit justification |
Why is this case informative for the phenomenon or proposition being studied? |
| Ethnographic research |
Settings, groups, activities, participants, and events may be sampled in relation to the cultural or social phenomenon |
Where and with whom can the relevant practices be observed and understood? |
| Qualitative descriptive or interview study |
Purposive selection may seek relevant experience, breadth, variation, or particular stakeholder perspectives |
Whose accounts are needed to answer the research question adequately? |
The labels and practices vary across qualitative traditions, so researchers should follow the methodological literature appropriate to the design they actually claim to use.
Do Qualitative Samples Need to Be Representative?
Not necessarily in the statistical sense.
A qualitative sample is often selected to illuminate experiences, meanings, processes, cases, interactions, or developing theoretical concepts rather than to estimate the numerical distribution of those characteristics in a population.
This does not mean that sampling quality is irrelevant. A poorly conceived sample may omit perspectives essential to the phenomenon, contain participants with insufficient relevant experience, or reflect nothing more than whoever was easiest to recruit.
The appropriate standard is therefore not automatically “Does this sample statistically represent the population?” but rather “Does this sampling strategy provide the information needed for this qualitative inquiry, and are its boundaries transparent?”
How Many Participants Should You Plan to Recruit?
Qualitative sampling does not have one universal participant number.
Sample adequacy depends on the study aim, methodology, specificity of the sample, richness and quality of the data, heterogeneity relevant to the analysis, and analytical strategy. The concept of information power proposes that a sample may require fewer participants when it contains more information relevant to the study. Its original formulation highlights the study aim, sample specificity, use of established theory, quality of dialogue, and analysis strategy as relevant considerations.
Saturation is also frequently invoked, but the term has multiple meanings and should not be treated as a ritual phrase meaning simply “no new themes emerged.” Researchers need to specify what form of saturation they mean and how they assessed it when saturation is part of their methodological logic.
The separate question of qualitative sample size, saturation, and information power therefore deserves attention after the sampling purpose and strategy have been established.
Recruitment Is Not the Same as Sampling
Sampling specifies which kinds of participants or cases you need and why. Recruitment concerns how you actually approach and enroll them.
You might purposively decide that you need instructors who have experienced three contrasting institutional AI-policy environments. You could then recruit those instructors through professional associations, institutional contacts, public invitations with screening, or referrals.
Reporting only the recruitment channel can obscure the sampling logic. Conversely, claiming purposive sampling without explaining how participants were located and selected leaves the procedure incomplete.
Report the Sampling Process Transparently
Qualitative readers need enough information to judge why the sample makes sense for the study.
The EQUATOR Network identifies reporting standards specifically for qualitative research, including the Consolidated Criteria for Reporting Qualitative Research, or COREQ, for interviews and focus groups. Transparent reporting should make clear who was eligible, how participants were selected and approached, the setting, relevant sample characteristics, and other methodological details needed to interpret the evidence.
If sampling evolved during the study, explain that evolution rather than presenting the final sample as though every decision had been fixed before the first interview.
