Maximum Variation vs. Homogeneous Sampling: Which Fits Your Qualitative Research Question?

Should Your Qualitative Participants Be Different from One Another or Share Similar Characteristics?

You are investigating how university teachers experience the integration of generative AI into student assessment. You could recruit teachers from different disciplines, institutions, career stages, and levels of AI experience. Alternatively, you could focus exclusively on first-year teachers who have recently introduced AI-generated feedback.

The first approach may reveal how experiences differ across contexts. The second may allow a more concentrated investigation of a particular group's experiences.

Which would produce stronger qualitative research?

Researchers sometimes assume that a more diverse sample is inherently more rigorous because it includes more perspectives. Others favor homogeneous samples because participants share characteristics that make their experiences easier to compare.

Neither assumption is sufficient. Diversity and similarity are not competing indicators of quality. They are sampling decisions that serve different analytical purposes.

The methodological question is whether the research seeks to understand variation across meaningfully different cases or develop a detailed interpretation within a relatively bounded group.

Choose Variation for Meaningful Contrasts and Homogeneity for Focused Depth

How Maximum Variation and Homogeneous Sampling Serve Different Analytical Purposes

What Is Maximum Variation Sampling?

Maximum variation sampling is a purposive strategy in which researchers deliberately select cases differing across characteristics considered important to the research question.

These differences may concern institutional context, professional experience, implementation conditions, participant roles, or other analytically relevant dimensions.

For example, a study examining institutional responses to generative AI might include teachers from universities with permissive, restrictive, and developing AI assessment policies.

The objective is not simply to collect as many different people as possible. It is to investigate how the phenomenon is experienced or organized under meaningfully contrasting conditions.

Maximum variation sampling may help researchers identify patterns that recur across different cases and differences that appear related to contextual circumstances.

However, observing a pattern across diverse purposively selected cases does not establish how frequently that pattern occurs in the wider population.

What Is Homogeneous Sampling?

Homogeneous sampling is a purposive strategy that deliberately selects cases sharing characteristics relevant to a focused research question.

For example, researchers might interview newly appointed university instructors who have used AI-generated formative feedback for the first time during the same academic year.

The shared characteristics help establish a relatively specific context for examining experiences.

Homogeneity does not mean that participants are identical or that their accounts will necessarily agree.

Participants sharing the same professional role or experience may interpret events differently, encounter different constraints, and respond in contrasting ways.

The purpose is to reduce variation on selected contextual dimensions so the analysis can concentrate on the phenomenon within a defined group.

What Is the Essential Difference Between the Two Strategies?

Both approaches use researcher judgment to select relevant cases. Their difference lies in which characteristics are deliberately varied and which are held relatively constant.

The choice should be justified through the research question rather than an assumption that diversity or similarity is inherently superior.

When Is Maximum Variation Sampling Appropriate?

Maximum variation sampling is useful when contextual differences are expected to influence how a phenomenon is experienced or understood.

Suppose researchers ask how university instructors negotiate generative AI assessment policies across different institutional environments.

Institutional policy, disciplinary expectations, and available technical support may be important sources of variation.

Deliberately selecting instructors from contrasting contexts allows researchers to examine how these conditions appear in participants' accounts.

This strategy may be particularly valuable when the research seeks to understand the boundaries of a process or develop an interpretation that accounts for contrasting circumstances.

However, variation must be analytically purposeful. Recruiting participants from many institutions without identifying why their differences matter may produce breadth without sufficient explanatory value.

When Is Homogeneous Sampling Appropriate?

Homogeneous sampling is useful when the research question concerns a particular shared experience or a narrowly defined participant group.

For example, a phenomenological study might investigate the experiences of first-year teachers required to use an AI-supported assessment platform during their initial semester.

Including senior administrators, experienced teachers, and students could shift the inquiry away from the specific experience being investigated.

A homogeneous sample can support focused interpretation by keeping the participant group aligned with the phenomenon.

However, the researcher should not assume that every characteristic must be identical. Differences that emerge within the group may still be analytically important.

Homogeneity should be defined only on characteristics necessary to establish the study's focus.

Which Characteristics Should Be Varied or Held Constant?

The selection of characteristics should follow the research question and relevant theoretical or contextual reasoning.

For a study of AI-supported assessment, possible dimensions include institutional sector, academic discipline, teaching experience, assessment format, and level of AI implementation.

Not all of these dimensions need to determine recruitment.

Researchers should ask whether a characteristic plausibly shapes the phenomenon and whether comparing cases along that dimension contributes to the intended analysis.

The examples illustrate analytical choices rather than mandatory sampling rules.

Does Maximum Variation Sampling Mean Maximizing Every Possible Difference?

No. Attempting to maximize every available characteristic may produce an excessively heterogeneous sample.

For example, a study with 15 participants cannot necessarily examine meaningful differences across six disciplines, five institutional types, four career stages, and several AI implementation conditions.

Each additional dimension increases the complexity of the analytical comparisons.

Researchers should prioritize a manageable number of theoretically or substantively relevant dimensions.

Maximum variation refers to deliberate contrast along selected dimensions, not unrestricted demographic diversity.

The problem of excessive heterogeneity in qualitative sampling becomes important when the study cannot examine the differences it deliberately introduces.

Can a Homogeneous Sample Still Contain Meaningful Variation?

Yes. Homogeneity is relative to the characteristics defining the sample.

Suppose every participant is a first-year university instructor who has implemented AI-generated feedback.

These instructors may still differ in confidence, institutional support, teaching philosophy, and perceived outcomes.

Such differences may emerge during analysis even though participants share the defining eligibility characteristics.

A homogeneous sample does not require uniform experiences or interpretations.

Indeed, identifying differences within a shared context may be one of its most valuable analytical contributions.

How Do These Strategies Affect Analytical Depth?

Maximum variation sampling can broaden the range of contexts and experiences examined, but it may require additional cases or more extensive data collection to understand each contrasting condition adequately.

Homogeneous sampling can support focused analysis within a specific group, although it may leave important contextual alternatives unexplored.

The relationship between sampling and analytical depth is not determined by participant diversity alone.

Interview quality, contextual information, research design, and analytical approach also influence the adequacy of the evidence.

Researchers should avoid assuming that a homogeneous sample is necessarily deeper or that a heterogeneous sample is necessarily more comprehensive.

Both strategies require sufficient analytical attention to the cases selected.

How Does the Research Tradition Affect the Choice?

Different qualitative traditions may place different demands on sampling.

Some phenomenological approaches emphasize participants who share experience of a particular phenomenon, making a relatively homogeneous sample potentially appropriate.

Comparative case studies may deliberately select contrasting cases to examine contextual differences.

Ethnographic studies may select settings or participants according to their roles within a cultural or organizational context.

Grounded theory studies may begin with purposive recruitment and subsequently use theoretical sampling to refine emerging categories.

These are tendencies rather than universal rules. The chosen sampling strategy should be consistent with the particular methodological tradition and its analytical objectives.

Can Maximum Variation and Homogeneous Sampling Be Combined?

Yes. A study may use homogeneity within groups and maximum variation across groups.

For example, researchers might select three institutional contexts with contrasting AI assessment policies and recruit instructors with similar responsibilities within each institution.

The institutional contexts provide variation, while the comparable participant roles support more focused cross-case interpretation.

Alternatively, a study may begin with a homogeneous sample and later introduce contrasting cases when the analysis identifies relevant contextual differences.

Such combinations should be planned or documented carefully.

Researchers should explain which characteristics are intentionally shared, which are varied, and how these decisions serve the analysis.

How Do You Determine Sample Adequacy?

Neither strategy has a universal sample size requirement.

Malterud, Siersma, and Guassora's information power framework emphasizes that qualitative sample adequacy depends on the study aim, participant specificity, theoretical grounding, dialogue quality, and analytical strategy.

A narrowly focused homogeneous sample may provide substantial information for a specific question, but this does not guarantee that a small number of interviews will be sufficient.

A maximum variation sample may require enough cases to examine important contrasts, although the required number depends on the intended analytical comparisons.

Researchers should also distinguish code saturation from deeper meaning saturation, as discussed by Hennink, Kaiser, and Marconi.

The sample adequacy rationale should be compatible with the qualitative approach rather than based on a fixed participant count.

Can Either Strategy Support Statistical Representativeness?

Not through conventional probability-based inference.

Both are purposive strategies, and individual population inclusion probabilities are generally unknown.

A maximum variation sample may include contrasting contexts, but it does not establish how frequently those contexts occur in the population.

A homogeneous sample may provide detailed accounts of a particular group, but it does not automatically represent all members of that group.

Qualitative findings may support theoretical interpretation or contextual transferability when justified by the evidence.

Researchers should distinguish those forms of inference from statistical population estimation.

How Should the Sampling Choice Be Reported?

The methodology should identify the research question, the purposive strategy, and the characteristics used to select cases.

For maximum variation sampling, explain which dimensions were deliberately varied and why they matter.

For homogeneous sampling, explain which characteristics define the group and why narrowing the sample supports the research purpose.

Where strategies are combined, describe the selection logic at each level.

Researchers should also report recruitment constraints, sample adequacy considerations, and important perspectives or contexts not included.

Frameworks such as COREQ and the Standards for Reporting Qualitative Research provide useful guidance for transparent reporting.

Two Sampling Strategies for Studying AI Assessment Practices

Under the maximum variation approach, the researcher might select institutions with different AI policies and resource conditions, then recruit teachers who have implemented AI-generated feedback.

The analysis could examine how institutional support, policy restrictions, and disciplinary expectations influence participants' experiences.

Under the homogeneous approach, the researcher might recruit first-year teachers who used AI feedback for the first time during the same academic term.

The analysis could focus on uncertainty, professional adjustment, and how participants interpreted their initial implementation experiences.

Neither strategy is automatically more rigorous. The maximum variation design would be poorly aligned with a question requiring a detailed account of one narrowly defined experience, while the homogeneous design would provide limited evidence for comparing substantially different institutional contexts.

The defensible choice follows the intended interpretation rather than a preference for diversity or similarity.

Common Misunderstandings About Maximum Variation and Homogeneous Sampling

How to Choose Between Variation and Homogeneity

Begin by identifying whether the research question concerns differences across contexts or a shared experience within a particular group.

In the methodology section, explain what characteristics were intentionally varied or held constant and why those choices support the intended analysis.

Do not justify maximum variation solely by saying that diversity improves credibility, or homogeneous sampling solely by saying that participants are similar.

The methodological rationale should show how the selected configuration of cases makes the research question answerable.

Before Choosing Your Qualitative Sampling Strategy, Check the Analytical Fit

Frequently Asked Questions About Maximum Variation and Homogeneous Sampling

Choose the Configuration of Cases That Serves the Inquiry

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.