03 · What You Need to Know
How Diversity and Depth Contribute to Qualitative Sampling
What Does Diversity Mean in Qualitative Sampling?
Diversity refers to meaningful differences among selected cases, participants, settings, experiences, or other units of analysis.
These differences may concern institutional context, professional role, experience with a phenomenon, geographical setting, or other characteristics relevant to the research question.
For example, a study of generative AI assessment practices may include instructors working under permissive, restrictive, and developing institutional policies.
The purpose is to examine how experiences and practices differ across those conditions.
Qualitative diversity should not be confused with statistical representativeness. Deliberately recruiting participants from several categories does not establish the proportion of those categories in a wider population.
Its analytical value depends on whether the selected differences help researchers understand the phenomenon.
What Does Depth Mean in Qualitative Research?
Depth refers to the extent to which a study develops a detailed, contextualized, and analytically meaningful understanding of the phenomenon.
It may involve examining participants' interpretations, decision processes, contradictions, contextual conditions, and changes over time.
For example, a study of teachers' experiences with AI-generated feedback may investigate how instructors initially adopted the technology, encountered implementation difficulties, revised their practices, and interpreted the consequences.
Depth is not determined solely by interview duration, transcript length, or the number of quotations included in a report.
It depends on the relationship between data quality, analytical engagement, contextual understanding, and the study's research question.
A small sample does not automatically produce deep analysis, just as a large sample does not automatically produce superficial analysis.
Why Are Diversity and Depth Sometimes in Tension?
Every additional participant category or contextual dimension may introduce new analytical questions.
Suppose a researcher interviews 20 instructors across ten universities, several academic disciplines, and substantially different AI implementation conditions.
The sample includes considerable contextual variation, but the researcher may have only limited evidence about each institutional environment.
If institutional context is central to the explanation, a small number of interviews per setting may be insufficient to understand relevant processes.
Alternatively, interviewing 20 instructors from one tightly defined group may provide a more focused account but offer limited evidence about experiences elsewhere.
The tension concerns the analytical work required by the sampling design, not a universal numerical relationship between diversity and depth.
When Should Diversity Be Prioritized?
Diversity is particularly useful when the research question asks how a phenomenon differs across contexts, participant roles, or experiences.
For example, researchers may investigate how university instructors negotiate AI assessment policies in institutions with different governance arrangements.
Institutional policy, technical support, and disciplinary expectations may be relevant dimensions of variation.
Deliberately selecting contrasting cases can help researchers examine which processes appear across contexts and which seem associated with particular conditions.
However, the differences should be chosen for substantive reasons.
Recruiting participants from every available demographic category may add complexity without improving the explanation.
When Should Depth Be Prioritized?
Depth may be particularly important when the research question concerns a specific lived experience, meaning-making process, or bounded institutional phenomenon.
For example, a study might examine how first-year instructors experience the requirement to use AI-generated formative feedback during their initial teaching semester.
A relatively focused participant group may allow the researcher to investigate common circumstances and differences within that experience.
Depth may also require repeated interviews, observations, document analysis, or extended engagement rather than simply recruiting more participants.
However, narrowing the sample should not eliminate contextual differences that are necessary for understanding the phenomenon.
The justification must establish why the chosen boundaries support the intended interpretation.
How Do Maximum Variation and Homogeneous Sampling Relate to This Decision?
Maximum variation and homogeneous sampling provide two purposive strategies that can help operationalize different analytical priorities.
Maximum variation sampling
Deliberately selects cases differing on dimensions relevant to understanding the phenomenon across contexts.
Homogeneous sampling
Deliberately selects cases sharing characteristics that define a focused analytical group.
Maximum variation may support contextual breadth, while homogeneous sampling may help concentrate analysis on a shared experience.
However, maximum variation does not automatically prevent depth, and homogeneous sampling does not guarantee it.
The researcher must still collect and analyze evidence appropriate to the intended interpretation.
Does Greater Participant Diversity Automatically Improve Credibility?
No. Diversity can strengthen a study when contrasting cases reveal conditions, perspectives, or experiences that would otherwise remain unexamined.
However, diversity that is unrelated to the research question may produce unnecessary analytical complexity.
For example, including instructors from ten disciplines may add little value if the study concerns a technical feature implemented identically across all participating settings and disciplinary context is not analytically relevant.
Conversely, disciplinary differences may be essential when assessment practices vary substantially across fields.
Researchers should therefore explain why each dimension of diversity matters rather than assuming that a more heterogeneous sample is inherently more rigorous.
Can a Homogeneous Sample Still Produce Contrasting Findings?
Yes. Participants can share defining characteristics while differing in their experiences and interpretations.
For example, all participants may be first-year university instructors who implemented AI-generated feedback during the same academic year.
Some may describe increased confidence and reduced workload, while others report uncertainty or additional verification demands.
These differences may be analytically important even though the sample is relatively homogeneous.
Homogeneity should therefore be understood in relation to specified selection characteristics, not as an expectation that participants will provide identical accounts.
How Can Researchers Combine Diversity and Depth?
A study may deliberately select contrasting settings while examining a relatively focused group within each setting.
For example, researchers might choose universities with different AI assessment policies and recruit instructors who have directly implemented those policies.
The institutional selection provides variation, while the participant criteria create a coherent basis for comparing experiences.
Alternatively, a researcher may begin with a focused sample and introduce additional contrasting cases when analysis reveals that particular contextual differences require examination.
The combination should be supported by a clear analytical rationale.
Researchers should avoid creating numerous subgroups unless the study can meaningfully examine them.
How Does the Unit of Analysis Affect the Balance?
The unit of analysis determines what the study is primarily trying to understand.
If the unit is an individual experience, depth may involve detailed accounts of how participants interpret a particular phenomenon.
If the unit is an institution, depth may require multiple data sources within each institution, while diversity may concern differences across institutional cases.
For example, comparing AI governance across four universities may require interviews with several relevant stakeholders and analysis of institutional documents within each case.
Recruiting one participant from each of 20 universities would provide broader institutional coverage but might not support the same level of institutional explanation.
The sampling strategy should reflect the unit of analysis rather than treating the number of individual interviewees as the only relevant measure of sample adequacy.
How Does Information Power Inform the Decision?
Malterud, Siersma, and Guassora proposed information power as a framework for considering sample size in qualitative interview studies.
The framework emphasizes the study aim, sample specificity, theoretical background, quality of dialogue, and analytical strategy.
A narrowly focused research question involving participants with highly specific experiences may require a different sample from a broad investigation involving substantially varied contexts.
However, information power does not provide a universal numerical formula or guarantee that a homogeneous sample is adequate.
Researchers should consider whether the selected cases and resulting data can support the intended analytical claims.
Where diversity introduces additional complexity, the sample and analytical procedures should be sufficient to examine that complexity.
What Is the Difference Between Code Saturation and Meaning Saturation?
Hennink, Kaiser, and Marconi distinguish code saturation from meaning saturation.
Code saturation concerns identifying the range of relevant issues or codes, while meaning saturation concerns developing a more comprehensive understanding of those issues.
A diverse sample may help reveal different topics or experiences, but identifying those topics does not necessarily establish sufficient interpretive depth.
Similarly, repeated accounts within a homogeneous sample may support detailed understanding without establishing that all relevant contextual perspectives have been considered.
Saturation is not defined or used identically across all qualitative traditions.
Researchers should therefore explain the form of adequacy relevant to their methodological approach rather than claiming saturation merely because interviews have become repetitive.
Can Too Much Diversity Weaken a Qualitative Study?
Yes, when the number or complexity of differences exceeds what the study can examine meaningfully.
For example, a small interview study spanning many institutions, participant roles, and implementation conditions may produce fragmented evidence about each context.
The researcher may identify interesting contrasts but lack sufficient information to explain them.
In such circumstances, narrowing the sampling dimensions or extending data collection may be appropriate.
The concern is not diversity itself but a mismatch between the sampling scope and analytical capacity.
The limitations of excessive heterogeneity in qualitative sampling should be considered during study design.
Can Too Much Homogeneity Weaken a Qualitative Study?
Yes. A narrowly defined sample may exclude experiences or conditions essential to the research question.
Suppose a study seeks to understand institutional barriers to AI-supported teaching but recruits only enthusiastic instructors from a well-resourced university.
The sample may provide detailed accounts of successful implementation while overlooking barriers encountered elsewhere.
If those excluded conditions are central to the phenomenon, the sample may be too narrow.
Researchers should distinguish a justified focus from an unnecessarily restrictive selection strategy.
Should Diversity or Depth Determine the Number of Participants?
Neither priority establishes a universal participant count.
A study emphasizing variation may require enough cases to examine meaningful contrasts, while a focused study may require sufficient engagement to develop a detailed interpretation.
However, increasing participant numbers is not the only way to improve either dimension.
Repeated interviews, observations, document analysis, or more focused questioning may contribute to analytical depth.
Additional recruitment settings or carefully selected contrasting cases may contribute to diversity.
The sample size rationale should therefore explain the evidence required rather than assume that diversity necessarily demands a large sample or depth necessarily demands a small one.
How Should the Balance Be Justified in a Research Paper?
The methodology should explain the research question, unit of analysis, purposive sampling strategy, and characteristics intentionally varied or held relatively constant.
Researchers should identify the analytical contribution expected from the selected cases and explain how sample adequacy was evaluated.
Where important contexts or perspectives remain outside the study, their implications should be discussed.
Reporting frameworks such as COREQ and the Standards for Reporting Qualitative Research can support transparent descriptions of qualitative sampling decisions.
The justification should demonstrate why the selected balance between diversity and depth is appropriate rather than claim that either characteristic automatically establishes rigor.
