Data Saturation in Qualitative Research: How Many Interviews Are Enough and How to Justify Your Sample

Sample size is straightforward to describe in quantitative research when a statistical power calculation can determine how many observations are required under specified assumptions. Qualitative research works differently. Researchers conducting interviews are usually interested in obtaining sufficiently rich and relevant information to understand experiences, meanings, processes, or perspectives rather than estimating a population parameter from a statistically representative sample.

This difference creates one of the most common questions in qualitative dissertations: how many interviews are enough? Data saturation in qualitative research is frequently used to answer that question. Saturation broadly concerns the point at which additional data collection is no longer producing sufficiently new or analytically important information for the purpose of the study. However, saturation is more complex than simply interviewing people until participants begin repeating the same words or ideas.

There is no universal number such as 10, 15, 20, or 30 interviews that automatically establishes saturation. Sample adequacy depends on the study aim, participant diversity, sampling strategy, interview quality, analytical approach, complexity of the phenomenon, and what the researcher means by saturation. This guide explains these issues in practical terms and shows how students can plan, assess, document, and justify qualitative interview sample size without relying on arbitrary numerical rules.

What Is Data Saturation in Qualitative Research?

Data saturation generally refers to a stage at which continued data collection produces diminishing amounts of new information relevant to the analytical purpose of the study. In interview research, researchers may notice that later interviews repeatedly support ideas already identified and contribute progressively fewer new codes, dimensions, or interpretations. At that stage, further interviews may add limited analytical value.

The concept is widely used but not defined consistently across qualitative traditions. Some researchers focus on whether new codes are still appearing, others on whether themes have been sufficiently developed, and grounded theory uses saturation in relation to the development of theoretical categories. Consequently, a dissertation should explain what saturation means within its particular methodology rather than simply stating that "saturation was reached."

This methodological alignment is important because qualitative research encompasses many different designs. Epic Essay's guide to thesis research methods explains that qualitative approaches are particularly useful for investigating experiences, meanings, and perspectives through methods such as interviews and observations. The criteria for adequate data should therefore reflect the purpose and analytical logic of the selected qualitative approach.

Saturation is not a predetermined number of interviews. It is a methodological judgement about whether the collected data are sufficiently developed for the study's analytical purpose.

Why There Is No Universal Number of Interviews for Saturation

A frequent mistake is searching for a single number that can be inserted into every qualitative proposal. This is attractive because a statement such as "20 interviews are required for saturation" appears precise and easy to defend. The problem is that two studies containing 20 interviews can differ enormously in participant diversity, interview depth, research scope, analytical method, and the complexity of the phenomenon being studied.

Consider a narrowly focused study involving employees who perform the same specialist role in one organisation. Participants may share many contextual characteristics, and detailed interviews may generate concentrated information relevant to a specific research question. Another study investigating experiences across several professions, countries, age groups, and organisational settings may require substantially broader data because meaningful variation exists across the sample.

Interview quality also matters. Twenty superficial ten-minute interviews do not necessarily contain more useful qualitative information than twelve carefully conducted hour-long interviews involving strong probing and detailed reflection. Counting participants without considering the information generated can therefore create a misleading impression of methodological rigour.

Do not claim that a particular number of interviews automatically proves saturation. The number should be accompanied by a methodological explanation of why the collected data were adequate for the research question and analytical approach.

Code Saturation vs Meaning Saturation

One of the most useful distinctions for students is between code saturation and meaning saturation. Research by Hennink and colleagues distinguishes the point at which researchers stop identifying substantially new issues from the point at which they have developed a rich understanding of those issues. These are not necessarily reached at the same time.

Code saturation concerns breadth. It is reached when additional interviews generate few or no new codes or thematic issues and the coding framework begins to stabilise. In Hennink and colleagues' methodological study of 25 in-depth interviews, code saturation occurred at around nine interviews in that particular dataset.

Meaning saturation concerns depth. Researchers continue examining whether further interviews add dimensions, nuances, variations, explanations, or contextual understanding to issues that have already been identified. In the same study, meaning saturation required approximately 16 to 24 interviews, demonstrating why identifying the main topics is not equivalent to understanding those topics fully.

The distinction is summarised below.

The numerical findings from one methodological study should not be transformed into a rule that nine interviews always achieve code saturation or that 24 always achieve meaning saturation. Their importance lies in demonstrating that different forms of saturation can require different amounts of data.

Data Saturation, Thematic Saturation, and Theoretical Saturation

Qualitative literature uses several related saturation terms, and they should not automatically be treated as synonyms. Data saturation is often used broadly to describe a point at which additional data become repetitive or cease to provide important new information. Thematic saturation commonly refers to the stage at which further analysis is no longer generating new themes or relevant thematic content.

Theoretical saturation has a more specific connection with grounded theory. In grounded theory, data collection and analysis proceed iteratively while emerging concepts influence subsequent sampling. Saturation concerns the development of theoretical categories and their relationships rather than merely discovering that participants are repeating similar statements.

This distinction matters when writing a dissertation methodology. A thematic analysis study should not automatically claim "theoretical saturation" simply because the phrase sounds more rigorous. The terminology should correspond to the actual methodology and analytical process used.

The table shows why the sentence "data collection continued until saturation" is incomplete. Readers need to know what form of saturation was relevant, how it was assessed, and what evidence led the researcher to decide that sufficient data had been collected.

What Factors Determine How Many Interviews You Need?

Qualitative sample adequacy depends on the information generated by the study rather than participant count alone. A narrow research question, relatively specific participant group, focused interview guide, and rich interviews may require fewer participants than a broad study involving substantial heterogeneity and several analytical comparisons.

The complexity of the phenomenon also matters. A study asking how a highly specific professional group uses one software system may produce concentrated accounts relatively quickly. A study examining identity, migration, discrimination, family relationships, and employment across culturally diverse participants is likely to contain more dimensions that require exploration.

The main factors include:

• The breadth or specificity of the research aim
• The heterogeneity of the participant population
• The sampling strategy
• The complexity of the phenomenon
• The depth and duration of interviews
• The interviewer's ability to obtain detailed accounts
• The quality and relevance of participant responses
• The analytical method being used
• Whether subgroup comparisons are required
• The type and depth of saturation being sought

These factors interact rather than operating independently. A diverse sample combined with a broad research question and comparative analysis may require substantially more interviews than a narrow study involving information-rich participants selected for highly specific experience.

How Participant Diversity Affects Data Saturation

Participant heterogeneity can substantially affect how quickly saturation develops. When participants share important characteristics and experiences relevant to the research question, later interviews may begin reinforcing existing patterns relatively early. A highly diverse population is more likely to generate different experiences, meanings, and contextual influences.

This does not mean researchers should make their samples artificially homogeneous merely to reach saturation faster. Sampling decisions should follow the research question. If differences between first-year and final-year students are important to the study, both groups need adequate representation even if this increases the amount of data required.

Subgroups also complicate saturation claims. A researcher may conduct 20 interviews overall but have only four participants from one important subgroup. Claiming saturation across the entire study may conceal limited evidence about that subgroup. Where comparisons are central, sample adequacy may need to be considered within analytically important groups rather than solely across the total sample.

Epic Essay's guide to dissertation methodology explains that qualitative studies commonly use purposive or theoretical sampling and that researchers must justify participant selection. Saturation should therefore be interpreted alongside the logic used to select participants.

What Is Information Power in Qualitative Research?

Information power provides another way to think about qualitative sample adequacy. Malterud, Siersma, and Guassora proposed that the more relevant information a sample contains for the study, the fewer participants may be needed. Instead of asking only when repetition occurs, researchers consider how much useful information the sample contributes to the research aim.

The framework identifies five important dimensions: the specificity of the study aim, sample specificity, use of established theory, quality of dialogue, and analysis strategy. A narrow aim, highly specific sample, strong interview dialogue, established theoretical framework, and focused analysis can increase information power. Broader aims and less specific samples may require more participants.

The relationship can be represented conceptually as follows.

Information power is especially useful during proposal development because researchers are often required to provide an expected sample before interviews have begun. Rather than pretending to know exactly when saturation will occur, the proposal can justify an anticipated range using the study's information needs and explain that final adequacy will be reviewed during data collection and analysis.

How Many Interviews Are Usually Enough for Qualitative Research?

Published methodological studies can provide useful reference points, but they should be treated as empirical examples rather than universal rules. Hennink and colleagues found code saturation around nine interviews in one study and required approximately 16 to 24 interviews for meaning saturation. Other research has found different numbers depending on the study design, population, analytical objective, and definition of saturation.

The important lesson is not that every student should conduct between 16 and 24 interviews. A tightly focused study involving a specific participant group may obtain sufficient information with fewer interviews, whereas a multi-site project involving several participant categories may require considerably more. The expected number should be justified from the methodological characteristics of the project.

Researchers should also distinguish planning from final determination. A dissertation proposal may state that approximately 15 to 20 interviews are anticipated based on the focused research aim, purposive sample, expected interview depth, and relevant methodological literature. During the actual study, concurrent review may show that additional interviews are required because important new dimensions continue to emerge.

Use published interview numbers as methodological evidence and planning benchmarks, not as automatic stopping rules.

Can You Decide the Number of Interviews Before Data Collection?

Universities and ethics committees often require students to specify an expected sample before recruitment begins. This can appear to conflict with saturation because researchers cannot know with certainty when sufficient information will emerge before conducting the interviews. The solution is to distinguish an anticipated sample from the final sample.

A proposal can provide a justified target or range based on the study aim, participant specificity, interview depth, analytical approach, feasibility, and previous methodological research. The researcher can then explain how adequacy will be evaluated as data collection and analysis progress. This is more transparent than claiming that saturation is already known to occur at a fixed number.

For example, a proposal might anticipate approximately 15 to 20 interviews with a narrowly defined participant group. It could explain that data will be reviewed iteratively and recruitment may continue if later interviews continue contributing substantively new codes, dimensions, or interpretations relevant to the research questions.

Any flexibility must remain consistent with ethics approval and institutional procedures. If recruitment beyond the approved range requires an amendment, the researcher should follow the relevant university requirements rather than simply continuing indefinitely.

Why Data Collection and Analysis Should Overlap

Saturation is difficult to assess if all interviews are completed before the researcher begins examining the data. Researchers need some understanding of what earlier interviews have contributed before they can determine whether later interviews are adding meaningful information. Iterative analysis therefore strengthens saturation decisions.

This does not necessarily mean completing a full final thematic analysis after every interview. Researchers can transcribe or review interviews progressively, maintain coding notes, update a preliminary coding framework, record new concepts, and compare later interviews with earlier material. The process provides evidence about whether analytical novelty is declining.

Epic Essay's guide to coding qualitative data explains that qualitative coding is iterative rather than strictly linear. That iterative logic supports saturation assessment because researchers can observe how the code structure develops as additional data are incorporated.

If analysis begins only after recruitment has closed, a statement that saturation determined the final sample becomes difficult to defend. The researcher could not have used saturation as a stopping criterion if the evidence needed to assess it had not yet been examined.

How to Assess Data Saturation During Interviews

Saturation assessment should be systematic enough that another reader can understand how the decision was reached. Researchers can track the appearance of new codes, note whether existing concepts acquire new dimensions, compare interviews across participant groups, and document when additional interviews mainly reinforce existing analytical patterns.

A simple saturation record can help. After each interview or small batch of interviews, the researcher records new codes, new dimensions of existing codes, contradictory cases, subgroup-specific insights, and questions requiring further exploration. This creates an audit trail rather than relying on memory.

A practical process can include:

1. Begin analysis while data collection is still underway.
2. Code interviews systematically using the chosen analytical approach.
3. Record genuinely new codes or concepts arising from each interview.
4. Distinguish new codes from new meanings within existing codes.
5. Track contradictory or deviant cases rather than ignoring them.
6. Compare patterns across important participant subgroups.
7. Review whether recent interviews mainly repeat existing material.
8. Conduct additional interviews when important concepts remain underdeveloped.
9. Document the reasoning used to decide that further data collection would add limited analytical value.

This process does not turn qualitative research into a mechanical counting exercise. Its purpose is to make the researcher's judgement visible and defensible while retaining the interpretive character of qualitative analysis.

How Thematic Analysis Relates to Saturation

Thematic analysis involves identifying and interpreting patterns of meaning across qualitative data. Saturation can sometimes be used within thematic studies, but researchers should specify exactly what they mean by it and ensure that the concept fits their chosen version of thematic analysis. Different qualitative traditions do not necessarily treat saturation identically.

Epic Essay's guide to thematic analysis explains that analysis moves from familiarisation and coding toward developing, reviewing, and defining themes. Early repetition of codes does not necessarily mean themes have been developed sufficiently because later interviews may deepen, complicate, or challenge the researcher's understanding.

This is where the distinction between code and meaning saturation becomes particularly useful. Researchers may reach a point where few new codes appear while still learning important things about how existing codes interact, differ across participants, or operate in different contexts. Ending data collection at the first sign of repetition can therefore produce analytically thin findings.

The saturation decision should reflect the depth of analysis required by the research question. If the dissertation aims to provide a nuanced account of how participants understand a complex experience, meaning and variation may matter more than merely demonstrating that the list of codes has stopped growing.

What Does Saturation Look Like in Practice?

Imagine a dissertation investigating how international master's students experience academic feedback. During the first interviews, the researcher identifies issues such as unclear comments, language difficulties, delayed feedback, uncertainty about expectations, emotional responses, and strategies for seeking clarification. Several new codes emerge quickly.

Later interviews may introduce few new topics but deepen existing ones. For example, "unclear feedback" might develop into different dimensions involving disciplinary terminology, indirect language, conflicting supervisor comments, and uncertainty about grading criteria. At this stage, code saturation may be approaching while meaning saturation has not yet been reached.

Eventually, additional interviews may mainly reinforce these dimensions without substantially extending their interpretation. Participants may provide different examples, but the examples fit within an already well-developed analytical account. If relevant participant variation has also been explored adequately, the researcher may have a stronger basis for judging the dataset sufficient.

The numbers in this example are illustrative only. Their purpose is to show that saturation develops through analysis rather than appearing automatically when a predetermined interview count is reached.

Can You Reach Saturation with Semi-Structured Interviews?

Semi-structured interviews are commonly used in studies where saturation or information adequacy is considered because they combine consistent topic coverage with flexible probing. Participants can address the same broad research areas while the interviewer explores unexpected experiences and asks for clarification or examples.

The quality of the interview guide therefore affects the information generated. Questions that are overly closed, leading, repetitive, or disconnected from the research questions may produce thin data even when many participants are interviewed. Rich interviewing depends on open questions, active listening, and appropriate probing.

Interview consistency also matters. If one participant receives detailed probing while another is asked only the headline questions, differences in data richness may reflect interviewing practice rather than genuine differences in experience. Training and reflective review can therefore strengthen both data quality and saturation assessment.

A strong saturation argument should consequently consider how the data were generated, not merely how many transcripts exist. Saturation based on shallow interviews is less convincing than adequacy supported by focused, information-rich conversations.

Common Mistakes When Claiming Data Saturation

One common mistake is writing that "saturation was reached after 15 interviews" without explaining what stopped emerging. The reader cannot determine whether the researcher means no new codes, no new themes, no new theoretical categories, or simply that responses felt repetitive. The type and assessment process should be stated.

A second mistake is claiming saturation retrospectively when all interviews were completed before analysis began. If saturation supposedly determined when recruitment stopped, there should be evidence that data were being reviewed during collection. Otherwise, the final number was determined by something else.

Another problem is treating repetition as sufficient regardless of depth. Participants may repeatedly mention "lack of support," but the researcher may not yet understand what support means, who is expected to provide it, when it is needed, or why its absence matters. Meaning saturation requires deeper analytical development.

Avoid using "saturation was achieved" as a ritual sentence with no evidence. Examiners need to see how saturation was defined, monitored, and connected to the final sample.

How to Report Data Saturation in a Dissertation Methodology

The methodology chapter should first explain the sampling strategy and expected sample rather than presenting saturation in isolation. Identify who was eligible, how participants were selected, why they were information-rich for the research question, and whether important subgroups needed representation. This establishes the context in which sample adequacy was judged.

The chapter should then define the form of saturation used and cite appropriate methodological literature. If the study assessed whether new codes and dimensions continued to emerge, explain how interviews were reviewed or coded during data collection. If information power informed the initial sample range, identify the factors that supported that reasoning.

A suitable write-up might state: "Participants were recruited purposively because they had direct experience of the phenomenon under investigation. An initial interview range was established using the focused study aim, sample specificity, anticipated interview depth, and methodological guidance on qualitative sample adequacy. Data collection and preliminary coding proceeded iteratively. Recruitment continued while interviews generated substantively new codes or dimensions of existing categories and ceased when subsequent interviews primarily reinforced the developed analytical framework without materially extending its interpretation."

The wording must match what actually occurred. Researchers should not claim concurrent coding, purposive sampling, meaning saturation, or information power if these processes were not genuinely used. Epic Essay's methodology guidance stresses the importance of explaining and justifying sampling and data-analysis procedures clearly rather than merely naming them.

How to Report Saturation in the Findings or Limitations

The methodology explains how adequacy was assessed, while later chapters can provide context about what the final sample achieved. If the final interviews generated little new analytical information, this can be reported concisely without overstating certainty. Researchers should avoid implying that saturation makes a qualitative sample statistically representative.

Limitations should acknowledge areas where information may remain uneven. For example, the overall dataset may appear well developed while one participant subgroup is relatively small. A transparent dissertation can recognise that limitation even if the broader analysis reached a defensible level of adequacy.

Researchers should also distinguish saturation from generalisability. Saturation concerns the adequacy of qualitative information for the analytical purpose of the study; it does not transform purposively selected participants into a probability sample representing an entire population. Claims should remain consistent with the qualitative design.

Clear reporting strengthens rather than weakens qualitative research. Explaining what the dataset can and cannot support allows readers to evaluate the credibility and transferability of the findings without relying on exaggerated claims.

A Step-by-Step Checklist for Deciding When You Have Enough Interviews

Determining sample adequacy is easier when researchers establish their logic before recruitment and then document how it develops during the study. The process should connect the research question, sampling strategy, interview quality, analytical method, and evidence of diminishing analytical novelty.

A practical checklist is:

1. Define the qualitative research aim and keep its scope realistic.
2. Select a sampling strategy appropriate to the research question.
3. Identify important participant characteristics and subgroups.
4. Decide what concept of saturation or sample adequacy fits the methodology.
5. Use relevant methodological literature to establish an anticipated sample or range.
6. Consider information power when planning the initial sample.
7. Conduct sufficiently detailed interviews using appropriate probing.
8. Begin reviewing and coding data while collection is still underway.
9. Track new codes, meanings, dimensions, contradictory cases, and subgroup differences.
10. Distinguish repetition of topics from genuine depth of understanding.
11. Continue recruitment when analytically important information remains underdeveloped.
12. Document the evidence supporting the final decision to stop data collection.

The checklist does not produce an automatic interview number. Instead, it provides a transparent process through which the final sample can be defended academically.

Justifying How Many Qualitative Interviews Are Enough

Data saturation in qualitative research is best treated as an analytical judgement supported by evidence rather than a numerical threshold. The appropriate number of interviews depends on the research aim, participant specificity, complexity of the phenomenon, quality of dialogue, sampling strategy, analytical approach, and depth of understanding required. For this reason, a number that is adequate for one dissertation may be inadequate or unnecessarily large for another.

Researchers should also be precise about what they mean by saturation. Code saturation concerns whether new issues continue to appear, while meaning saturation asks whether those issues have been understood in sufficient depth. Theoretical saturation belongs to a more specific grounded-theory logic, and information power offers a useful framework for thinking about sample adequacy from the relevance and richness of the information available.

The strongest dissertation therefore does not claim that a fixed number of interviews is universally enough. It provides a justified anticipated sample, analyses data iteratively where the methodology requires it, documents how new information develops, and explains why the final dataset was adequate for answering the research questions. That transparent reasoning is more defensible than any arbitrary interview count.