🔍💡 Interview sample: stop chasing a “good number” and learn to justify your choices (opening, focusing, closing) based on the field. #QualitativeResearch🧩📊 562 studies scrutinised: the strength of a qualitative inquiry does not lie in the magic number of interviews, but in how you construct and explain your sample. #PublicHealth
📌 This text is directly useful for any professional or doctoral student conducting qualitative interviews (public health, prevention, medico-social, social work) who needs to justify the size of their sample before an ethics committee, a funder, or a review. It shows, based on 562 studies, that there is no none universal “good number” and that the quality of the methodological argumentation matters more than the raw volume of interviews. It proposes a simple framework (opening, focusing, closing movements) to decide, document, and defend an appropriate number of interviews for each project
Source: ✍️ How Many Interviews Do I Need? An Examination of Interview Numbers and Sampling Moves in Qualitative Research -Kasper Trolle Elmholdt, Michael Gill, Jeppe Agger Nielsen – journal Organizational Research Methods (SAGE) - 2026
📜🔗LINK to the source
Number of pages: 36
1. Analytical summary
1.1. Context: the false quest for the “right” number of interviews
The authors start from a widely shared observation: doctoral students, research teams, and evaluators constantly ask, "how many interviews are needed?", with answers often being normative (12, 30, 50, 100…) and poorly related to the actual analytical aims of the studies. Their review covers 562 qualitative articles based on interviews, published between 2010 and 2019 in six major management journals (AMJ, ASQ, Organization Science, SMJ, Journal of Management Studies, Organization Studies). Sample sizes range from 2 to 682 interviews, with a median of 46 and an average of 60, and moderate differences between North American journals (average 74) and European journals (average 48). The authors show that there is no strong link between the number of interviews and the methodological approaches (type of case study, grounded theory, ethnography, duration, number of sites, combination with observations or documents). Most importantly, 84% of the articles do not explicitly justify their sample size, or merely provide a vague reference to “saturation” without specifying how it was assessed. This situation is analysed as the result of a mix of political pressures (journal expectations, “publish or perish”) and field pragmatism (access, resources, opportunities), which disconnect sampling practices from the stated methodological foundations.
1.2. Operational contributions: the “sampling moves” to think about and document the sample
By analysing in detail 60 exemplary articles, the authors identify ten "sampling movements" grouped into three families: opening movements (understanding the context, refining the question, adjusting the guide), focusing movements (stratifying, bounding, cleaning, decomposing the sample), and closing movements (validating interpretations, deciding on the stopping point, clarifying the limits). These movements describe what a research team actually does to move from an initial idea to a credible sample, rather than aligning with a predetermined number. The text provides very concrete examples (pilot interviews, stratification by function or seniority, selection of rich interviews to avoid "data asphyxiation", use of "member check" interviews, stopping at saturation or time/funding constraints) and links them to different epistemological traditions (realist, constructivist, discursive). At the end of the article, the authors synthesise this framework into a series of practical questions for researchers, evaluators, and committees (how did we open the field? how do we justify the inclusion/exclusion of certain groups? why did we stop here?). The central issue is to shift the discussion from "how many interviews?" to "what movements have we made to produce sufficient and relevant data for our question?".
2. Key points of the document
No "good" range of interviews emerges, even in the most prestigious journals (pp. 9–12).
Out of 562 studies, sample sizes range from 2 to 682 interviews, with a median of 46 and an average of 60, and 18% of studies having more than 100 interviews while 15% have 20 or fewer. This dispersion persists even after excluding extreme values, allowing the authors to refute the idea of an expected threshold for publishing in high-level journals.
Design characteristics (study type, number of sites, level of analysis, duration, combined methods) only weakly explain the number of interviews (p. 12–13, table 2).
Multi-level studies (average 64 interviews) do not systematically conduct more interviews than single-level studies (average 54), and designs such as grounded theory, ethnography, case study, or process study all have similar medians (44 to 60). Even the use of other methods (observations, documents) does not systematically result in smaller or larger samples.
The justification for sample size is often absent or reduced to a superficial reference to saturation (p. 12–14).
84% of articles do not specify how the number of interviews was determined; only 15% explicitly mention data or theory saturation, without describing how it was observed (repetition of themes, absence of new categories, etc.). The authors highlight that this 'silencing' of sampling decisions contributes to a misleading objectification of qualitative research, where the total number replaces reflection on the process.
The framework of 'sampling moves' structures the entire cycle of qualitative inquiry (p. 14–23, tables 3–4).
Opening movements include the use of exploratory interviews to understand the context, identify key actors, refine the research question, and stabilise the interview guide. Focusing movements include stratification (by role, gender, seniority, site), bounding the sample to address access constraints, curating a subset of rich interviews for in-depth analysis, and detailed decomposition of the sample (by phases, levels, groups) in the report. Closing movements cover member checks, setting a stopping point (often in the form of theoretical saturation or structural coherence of narratives), and clarifying the limits imposed by time, funding, or refusal to participate.
The authors propose guiding questions to assess the quality of sampling decisions, rather than a new 'recipe' (p. 24–26, table 5).
For the opening, they invite documentation on how the initial interviews served to understand the context, select cases, and categorise preliminary interviews. For the focusing, they suggest explaining why certain groups were included or excluded, how empirical exposure was maximised, and why some interviews are more mobilised than others in the analysis. For the closing, they emphasise the need to clarify stopping criteria (saturation, internal coherence, constraints) and to clearly articulate the final sample size to the epistemological and ontological positions of the team.
3. Action points for local actors
Build a sampling plan by 'movements' rather than by target number (p. 14–18, 24–26).
For a public health or medico-social project, formalise three phases from the protocol:
an opening phase (2–5 exploratory interviews) to understand the local context, refine the question, and adjust the interview guide;
a focusing phase where relevant strata (users, close caregivers, professionals, decision-makers, associations, etc.) and criteria for variation (age, gender, territory, type of structure) are defined;
a closing phase where verification interviews (users, association representatives) are planned and where the stopping criteria (thematic saturation, coherence of narratives, scheduling constraints) are clarified. This structuring can be integrated into the protocols submitted to the CPP, internal ethics committees, or project calls.
Systematically document access constraints and their effects on the sample (pp. 19–22, 23).
In sensitive contexts (violence, addictions, HIV, mental health, extreme precarity), access difficulties or refusals should be considered as data and not as 'failures'. It is recommended to record, in the methods section, the refusals, the intermediaries mobilised, the compromises (for example, using intermediary professionals rather than the users themselves) and the target adjustments (fewer institutional leaders, more field social workers, etc.). This allows for anticipating criticisms regarding the 'small size' of the sample and demonstrates how these constraints also signal power relations or issues of stigma in the field.
Establish a reasoned curation of interviews to preserve analytical depth (pp. 19–21).
When many interviews are conducted (for example in regional or national frameworks), it is useful to clearly announce which interviews will be centrally mobilised for detailed analysis ("rich" interviews, complete trajectories, diversity of profiles) and which will serve as more occasional support. Curation, if explicit, helps to avoid "suffocating" the analysis with an unmanageable volume of data, while ensuring sufficient coverage of variations in context or situation.
Plan validation interviews with different types of stakeholders (pp. 21–22).
In prevention or care organisation projects, additional targeted interviews can be scheduled at the end of the analysis to discuss the results with users, peers, frontline professionals, or association leaders. These "member checks" enhance the credibility of interpretations, help identify blind spots (e.g. invisibilised groups, unexpected effects of a framework), and provide a strong argument to decision-makers to show that the conclusions are not merely "academic".
Explicitly align the sample size with the epistemological stance and expected uses (pp. 6–7, 22–24, 26).
For studies aiming to produce detailed narratives (narrative, discursive, ethnographic approach), assume a limited number of interviews by explaining that the goal is coherence of meaning rather than statistical coverage. For more inductive work such as grounded theory or comparative case studies, clarify how saturation (or structural coherence) has been assessed, and how the number of interviews is sufficient to support the intended organisational or public policy recommendations.
4. Additional References
Guetterman TC. “Deciding the Sample Size in Qualitative Interview Studies: A Systematic Approach.” (2023 or later)
Methodology article proposing a decision-making framework to size and justify the sample sizes of interviews, integrating objectives, expected heterogeneity, and resources. Verified URL:https://journals.sagepub.com/(search by title/author required).
NCRM (UK). “Sample Size in Qualitative Research – Methodological Guidance” (updated after 2023).
Practical guidance, very useful for funding applications and protocols, discussing saturation, variation, feasibility, and justification to ethics committees. Verified URL:https://www.ncrm.ac.uk/resources/online/.
Wutich A., Beresford M., Bernard HR. “Sample sizes for 10 types of qualitative data analysis”International Journal of Qualitative Methods2024 [already cited by Elmholdt et al.].
Proposes orders of magnitude for sample size for different types of analysis (thematic, discursive, ethnographic, etc.), complementing the framework of “sampling moves” to calibrate and justify the field strategy. Article accessible via the journal's website:https://journals.sagepub.com/home/ijq.
5. Frequently Asked Questions (FAQ – 7 questions/answers)
Q1 – Does this text say how many interviews I should conduct for a “good” qualitative study? (pp. 1–5, 9–12)
A1 – No. The authors show, based on 562 studies, that there is no “magic number” of interviews: for articles published in high-level journals, there are surveys with 2 interviews as well as those with over 600, without a clear minimum or maximum threshold emerging.
Q2 – How do the studies examined justify their sample size in practice? (pp. 12–14)
R2 – In 84% of cases, the justification is absent or very implicit; when a justification exists, it mainly invokes the notion of saturation but does not clarify the process (which themes, which indicators, which iterations). This leads the authors to propose the language of “sampling moves” to make these decisions more visible.
Q3 – Do grounded theory, ethnography, or case study research systematically require more interviews? (p. 6–7, 12–13)
R3 – No. The analyses show only weak variations between these designs, with fairly close medians (around 44–60 interviews), and no clear relationship between design type and sample size when controlling for extreme values.
Q4 – What exactly do the “opening moves” in sampling cover? (p. 14–18, tables 3–4)
R4 – These are the interviews used to: understand the context (exploratory, pilot), refine the research question (identify what really poses a problem for the actors), and stabilise the interview guide (adjustments after the first interviews). These moves may involve a few additional interviews, but they secure the relevance of the entire study.
Q5 – How to apply the “focusing moves” to a health or social survey? (p. 18–21)
R5 – This involves stratifying the sample (by type of actors, institutions, territories), bounding the sample considering access constraints, selecting a subset of particularly rich interviews for fine analysis, and precisely describing the composition of the sample (by phases, levels, groups) in the report.
Q6 – When do we know that we can stop conducting interviews? (p. 21–22, 25–26)
R6 – The authors list several criteria: theoretical saturation (no new themes), structural coherence of narratives (discourses that 'hold' together), validation by participants (member checks), or explicit constraints (access, time, funding). They recommend making these criteria transparent rather than referring solely to a number.
Q7 – How does this framework help ethics committees, funders, or reviewers to evaluate a protocol? (pp. 24–26)
R7 – Table 5 offers simple questions: how was the field opened? why is a particular group included or excluded? how have methodological challenges influenced the sample? why are the data considered sufficient at this stage? These questions can be incorporated into the review grids of the files to assess the quality of the approach beyond the number of interviews.
6. Rewriting in Easy-to-Read Language
6.1. Easy-to-Read Summary
Title:
How many interviews do I need?
Main idea:
Researchers often conduct interviews to understand people's lives at work or in organisations. Many believe that a 'good number' of interviews is needed for the study to be serious. The authors looked at 562 studies with interviews and saw very different numbers, from 2 to over 600 interviews. They say there is no perfect number of interviews. What matters is to clearly explain how the people to be interviewed are chosen and why the process stops at a certain point.
6.2. Key points in Easy-to-Read Language
1. Studies use very different numbers of interviews.
Some studies use very few interviews, for example, 2 or 13.
Other studies use a lot of interviews, more than 100.
There is no single good number that works for everything.
2. Researchers rarely explain why they conducted this number of interviews.
In most articles, the authors do not say how they decided on the number of interviews.
They sometimes write, “we reached saturation,” without explaining what that means.
3. The authors talk about three key moments in the interview process.
Opening:conducting a few interviews at the beginning to understand the situation and improve the questions.
Focusing:choosing which groups to see, how many people in each group, and which interviews to examine closely.
Closing:deciding when to stop the interviews and checking that the results are correct with the participants.
4. It is important to state what limits the number of interviews.
Sometimes, it is difficult to meet certain people (fear, lack of time, sensitive topics).
Sometimes, the project must stop due to time or money.
This should be clearly stated in the report.
5. Quality is more important than quantity.
A small number of well-conducted interviews can yield useful results.
Researchers must show how they asked the questions and how they analysed the responses.
7. Cross-sectional analysis – Values of Health Practices
Literacy:The document is not particularly accessible to audiences unfamiliar with the methodology, but it emphasises the need to clearly describe sampling decisions, which can enhance the understanding of professional readers.
Empowerment:The beneficiaries are not at the centre of the text, but the emphasis on member checks and validation with respondents paves the way for a more active involvement of the concerned individuals in the interpretation of the results.
Participation:Participation is mainly considered at the research level (exploratory interviews, verification of analyses), and not as a systematic co-construction of questions or protocols.
Community health:The text is situated in the field of management, but its proposals on how to document difficult access and field constraints are transferable to community health approaches where power relations and the invisibility of certain groups are strong.
Ethics:The authors invite making visible the access constraints, sample limits, and curation choices, which aligns with an ethics of transparency and reflexivity regarding possible biases.
Human rights:By emphasising the importance of explaining why certain groups are not present (refusals, institutional barriers), the text encourages reflection on the equity of access to voice in qualitative research.
Intersectorality:The study mobilises research involving businesses, NGOs, and administrations, which demonstrates the interest in samples that include different levels and types of actors, relevant for intersectoral approaches in health and social matters.
Partnership:The examples of member checks and triangulation with practitioners or 'elites' suggest forms of limited but real partnership between researchers and field actors.
Combating discrimination:The text highlights the importance of documenting the difficulties of access to stigmatized groups (for example, sexual minorities, banking sectors in crisis), which allows for the identification of structural discrimination in access to research and in the data produced.
8. Assessment of the reliability of the resource
Scientific relevance.
The article is published in a highly demanding international methodology journal and is based on a large corpus of 562 studies from six leading journals in management and organisations. The approach is transparent (inclusion criteria, multiple coding, detailed descriptions of distributions, tables 1–5) and mobilises a solid conceptual framework (saturation, ontology/epistemology, triangulation) while recognising the limitations of the corpus (discipline, period, pre-COVID).
Operational relevance.
For public health, prevention, and medico-social teams, the text provides an immediately reusable framework for designing interview plans, writing protocols, addressing criticisms regarding sample size, and training teams in methodological reflexivity. It does not provide a quantified 'recipe', but rather a framework for clarifying decisions, which is consistent with intervention research approaches, action research, and documented associative projects.