03 · What You Need to Know
What Saturation Means and When It Should Guide Interview Sampling
What Is Data Saturation in Qualitative Research?
Data saturation generally refers to a point in data collection and analysis at which additional material contributes little or no new information according to a specified analytical criterion.
However, this broad definition conceals important differences.
In some studies, researchers assess whether additional interviews generate new descriptive codes. In others, they consider whether the meanings and variations associated with identified themes are sufficiently understood.
Grounded theory researchers may instead evaluate whether developing theoretical categories and their relationships have been adequately elaborated.
These are not interchangeable stopping criteria.
Researchers should therefore identify the particular meaning of saturation relevant to their study rather than assuming that the term has one universally accepted operational definition.
Why Is Saturation Used to Determine When Interviews Should Stop?
Qualitative researchers often begin with an initial sampling plan while recognizing that the informational contribution of future interviews cannot be known completely in advance.
When analysis occurs alongside data collection, researchers can examine what the material already contributes and identify areas requiring further investigation.
Saturation provides one possible rationale for deciding when additional recruitment is unlikely to contribute substantially to a specified analytical objective.
Its appeal is understandable. Researchers need a defensible reason to stop collecting data without assuming that every possible participant must be interviewed.
However, saturation should not be used merely as a convenient label for the moment when recruitment becomes difficult, resources are exhausted, or the researcher has reached a preferred sample size.
Its methodological value depends on the relationship between the stopping criterion and the analysis being undertaken.
Code Saturation: When Few New Codes Are Identified
Code saturation concerns the identification of new issues or categories during analysis.
For example, a study of university teachers' experiences with generative AI might identify recurring codes relating to institutional policy, assessment concerns, technical limitations, and professional learning.
After additional interviews, the researcher may find that few new codes are being introduced.
This can indicate that the analysis has identified a substantial portion of the issues represented within the sampled material.
However, code saturation does not necessarily mean that the researcher understands each issue in sufficient depth.
A code such as "institutional uncertainty" may encompass very different experiences. Some teachers may lack clear guidance, others may encounter contradictory instructions, and others may disagree with established policy.
Recognizing the category is not equivalent to understanding its internal variation.
Meaning Saturation: When Themes Are Understood in Sufficient Depth
Meaning saturation concerns the development of a more comprehensive understanding of identified issues.
It involves examining nuances, contextual variation, relationships, and potentially contrasting accounts.
Hennink, Kaiser, and Marconi (2017) distinguished code saturation from meaning saturation in an empirical analysis of qualitative interviews.
Their work illustrates why identifying recurring thematic categories may require fewer interviews than developing a richer understanding of what those categories mean.
For example, researchers may quickly identify that teachers experience difficulties with institutional AI policies.
Further interviews might reveal how these difficulties differ between disciplines, how teachers negotiate conflicting expectations, and why some perceive policy restrictions as protective while others consider them obstructive.
Meaning saturation therefore requires more than counting newly identified codes.
The researcher must consider whether the available accounts adequately support the depth and complexity of the intended interpretation.
Theoretical Saturation: A Distinct Concept in Grounded Theory
Theoretical saturation is associated particularly with grounded theory traditions.
It concerns the development of theoretical categories, their properties, dimensions, and relationships through iterative data collection and analysis.
In grounded theory, researchers may select additional participants or situations because emerging analysis identifies a category requiring further development.
For example, an emerging theory about teachers' AI adoption might identify "negotiating institutional legitimacy" as an important process.
The researcher may seek additional cases that clarify how this process operates under different institutional conditions or how it relates to professional autonomy.
Theoretical saturation is not simply the absence of new interview topics. It concerns the adequacy of theoretical development within the particular grounded theory approach.
Researchers should avoid claiming theoretical saturation when their study does not undertake theoretical sampling or develop theoretical categories.
How Do Different Forms of Saturation Compare?
| Criterion |
Primary concern |
Possible indication |
Important limitation |
| Code saturation |
Identifying recurring issues or categories |
Additional interviews produce few new relevant codes |
Does not establish depth of understanding |
| Meaning saturation |
Understanding thematic depth and variation |
Additional accounts contribute little substantive nuance to established themes |
Depends on the analytical scope and interpretation |
| Theoretical saturation |
Developing theoretical categories and relationships |
Further theoretically directed sampling contributes little to category development |
Requires an appropriate grounded theory framework |
| Information power |
Adequacy of information relative to the study purpose |
The sample provides sufficiently relevant and detailed material for the intended analysis |
Is a framework for judgment, not a mechanical saturation test |
These approaches can inform sampling decisions, but their methodological assumptions differ. Researchers should not combine their terminology without explaining how the concepts are being used.
Is Saturation Appropriate for Every Qualitative Methodology?
No. The appropriateness of saturation depends on the study's methodological and epistemological commitments.
In some forms of qualitative description, thematic analysis, and grounded theory, saturation may provide a useful framework when it is defined consistently with the analytical approach.
However, Braun and Clarke (2021) have questioned the conceptual compatibility of saturation with reflexive thematic analysis, particularly where themes are understood as interpretive constructions developed through the researcher's engagement with data.
Under that approach, analysis does not simply involve discovering a finite collection of themes waiting to be exhausted.
Likewise, narrative inquiry and some phenomenological or ethnographic approaches may prioritize detailed cases, contextual understanding, or interpretive development rather than demonstrating that no further information could emerge.
Researchers should not impose saturation on a methodology merely because reviewers or supervisors expect to see the term.
The stopping rationale should follow the logic of the study.
Can Saturation Be Determined Before Data Collection?
Actual saturation cannot ordinarily be demonstrated before relevant data have been collected and analyzed.
Researchers can, however, plan how they intend to assess it.
A protocol might specify an initial recruitment target, a preliminary coding process, and a procedure for reviewing the informational contribution of successive interviews.
The researcher may also identify the level at which saturation will be assessed, such as the entire sample or analytically important participant groups.
Such planning is different from declaring that saturation will occur after a predetermined number of interviews.
Empirical studies of saturation can inform recruitment estimates, but they cannot guarantee the point at which a new study will reach its specified criterion.
The broader task of planning an initial interview sample size should therefore be distinguished from evaluating the achieved sample during analysis.
What Does Empirical Research Tell Us About Saturation?
Guest, Bunce, and Johnson (2006) examined the emergence of themes in interviews involving a relatively homogeneous participant group. Their study found that many commonly identified themes appeared within the first 12 interviews under the conditions examined.
Hennink, Kaiser, and Marconi (2017) demonstrated that code saturation and meaning saturation may occur at different stages of data collection.
Hennink and Kaiser (2022) reviewed empirical studies testing saturation and reported that studies with relatively homogeneous populations and narrowly focused objectives often observed saturation within approximately 9 to 17 interviews.
These findings provide useful context but should not be interpreted as universal requirements.
The point at which saturation is observed depends on the definition used, the population, sampling strategy, research question, and analytical procedure.
A study involving multiple institutions, heterogeneous participants, or complex theoretical development may have substantially different information requirements.
How Can Saturation Be Evaluated Systematically?
When saturation is methodologically appropriate, researchers should establish a transparent procedure for evaluating it.
One possible approach is to analyze interviews progressively and document whether additional interviews contribute new relevant codes or substantive developments in existing categories.
Researchers may use a saturation tracking matrix containing the interview identifier, participant characteristics, newly identified codes, and relevant changes to the analytical framework.
Guest, Namey, and Chen (2020) proposed a structured approach for assessing thematic saturation that considers a base set of interviews, subsequent runs of interviews, and the proportion of newly identified information.
Such procedures can improve transparency, but their parameters remain methodological choices requiring justification.
They are most suitable when the analytical approach treats the identification of codes or themes in a way compatible with the proposed assessment.
A numerical tracking procedure should not be mistaken for an objective proof that all possible meanings have been discovered.
Why Repetition Alone Is Not Enough
Participants may repeat similar statements because the interview guide consistently directs them toward the same topics.
For example, if every interview asks only about technical difficulties, repeated accounts of connectivity problems may reflect the narrowness of questioning rather than comprehensive understanding of technology adoption.
Repetition may also arise because participants share a common institutional context or because recruitment has excluded people with contrasting experiences.
Before treating repetition as evidence of saturation, examine whether the sampling strategy and questions have provided reasonable opportunities for relevant variation to emerge.
The researcher should also consider whether the analysis has developed sufficiently beyond descriptive labels.
Repeated codes can coexist with substantial unexplored differences in meaning.
What If New Themes Continue to Emerge?
The emergence of new material does not automatically mean that interviews must continue indefinitely.
Researchers should examine whether the new issue is relevant to the research question, whether it changes the developing interpretation, and whether it identifies an important gap in the sample.
Some unexpected issues may justify additional recruitment. Others may be peripheral to the study's defined scope.
For example, a participant might introduce an unrelated institutional concern that is interesting but not necessary for answering the research question.
The researcher should not expand the study without considering its conceptual boundaries, resources, and ethical approval.
Conversely, an unexpected account that challenges a central interpretation may warrant further investigation even if most interviews have become repetitive.
Analytical relevance matters more than simply counting every newly mentioned topic.
Should Saturation Be Assessed Separately for Participant Groups?
When a study intends to compare substantively different groups, overall saturation may conceal important gaps.
Suppose a researcher interviews teachers from institutions with restrictive and permissive AI policies.
If most interviews involve teachers from permissive environments, recurring themes across the overall dataset may not establish adequate understanding of experiences under restrictive policies.
Researchers should consider whether important groups or contexts require separate evaluation of analytical coverage.
However, dividing the sample into many small categories can make saturation assessments unstable or conceptually inappropriate.
The relevant units of comparison should follow the research question and analytical design.
How Does Interview Quality Affect Saturation?
Saturation assessments depend on the information generated during interviews.
If questions are unclear, leading, or insufficiently probing, participants may provide brief or repetitive accounts that fail to reveal relevant complexity.
The researcher may then mistakenly conclude that additional interviews have little to contribute.
Interview structure and probing therefore influence the credibility of saturation judgments.
The degree of flexibility in a semi-structured interview guide should permit relevant clarification without allowing the inquiry to lose its substantive focus.
Researchers should also consider whether interview questions impose assumptions from the conceptual framework, potentially restricting the range of experiences participants can describe.
Information Power as an Alternative to Saturation
Malterud, Siersma, and Guassora (2016) proposed information power as an alternative framework for reasoning about qualitative interview sample adequacy.
Rather than requiring researchers to demonstrate that no new information emerges, information power considers whether the sample provides sufficiently relevant information for the study's purpose.
The framework considers the study aim, sample specificity, established theory, dialogue quality, and analytical strategy.
For example, a narrowly focused study involving participants with extensive experience of a particular phenomenon may obtain substantial relevant information from a relatively small sample.
A broader study involving diverse contexts and cross-case comparisons may require more participants.
Information power is not a numerical formula or an automatic stopping rule. It supports a reasoned assessment of sample adequacy in relation to the intended analysis.
What Other Reasons Can Justify Ending Recruitment?
Some qualitative studies use fixed samples, bounded case selection, or analytical criteria other than saturation.
A narrative study may examine a deliberately selected set of individual accounts. An ethnographic investigation may define its data collection through field engagement, research scope, and contextual adequacy.
A qualitative evaluation may recruit participants across specified stakeholder categories according to a predetermined design.
These approaches can be defensible when the sample and data support the intended claims.
Practical constraints such as time, access, and funding also affect recruitment. They should be acknowledged transparently, but they do not automatically establish methodological adequacy.
If the achieved sample cannot support the original analytical objectives, the researcher may need to narrow the scope of interpretation or identify limitations.
How Should Saturation Be Reported in a Research Paper?
A statement such as "Interviews continued until data saturation was achieved" provides little information about how the decision was made.
More informative reporting explains the form of saturation assessed, the analytical procedure, the timing of the assessment, and the evidence supporting the stopping decision.
Researchers may report how interviews were coded, how new information was tracked, whether relevant participant groups were adequately covered, and why additional recruitment was considered unlikely to change the analysis materially.
If saturation was not used, explain the alternative sample-adequacy rationale.
Reporting frameworks such as COREQ and the Standards for Reporting Qualitative Research support transparent accounts of sampling and data collection, but they should not be interpreted as requiring saturation for every qualitative methodology.
The objective is to make the stopping decision understandable and open to methodological scrutiny.
