You've designed the interview guide. So naturally, you've got IRB approval. You're ready to talk to real humans Small thing, real impact..
Then the question hits: how many people do I actually need to talk to?
If you've ever stared at that question and felt the panic rise — you're not alone. Still, it's a concept. The answer isn't a number. Every qualitative researcher hits this wall. And most guides make it way more complicated than it needs to be Most people skip this — try not to. Still holds up..
What Is Sample Size in Qualitative Research
Sample size in qualitative work isn't about statistical power. It's not about representing a population so you can generalize to millions. It's about information power — a term Malterud and colleagues coined back in 2016 that actually makes sense That's the part that actually makes a difference..
Think of it this way: each participant adds a layer of understanding. The fifteenth might confirm patterns you're already seeing. The twentieth? The fifth adds nuance. So naturally, the first interview gives you a foundation. Often just noise.
But here's what trips people up: there's no magic number. Here's the thing — a study exploring how oncologists break bad news might need 12–15 interviews. Think about it: a study on the lived experience of rare disease caregivers might only need 6–8 because the phenomenon is so specific and the data so rich. Worth adding: a grounded theory study building a new theoretical framework? That could run 30, 40, even 50+ Easy to understand, harder to ignore..
The unit of analysis matters too. Day to day, observing clinic visits? Here's the thing — analyzing documents? Running focus groups? Are you interviewing individuals? Each "unit" yields different depth.
The saturation conversation
You'll hear "data saturation" thrown around like it's a finish line. Plus, it's not. Plus, saturation means you're no longer learning new things — no new codes, no new themes, no new variations on the story. But saturation is theoretical, not numerical. Think about it: you don't know you've hit it until you're there. And "there" shifts depending on your methodology.
Phenomenology chases depth. Also, you might interview 6 people three times each. Ethnography chases breadth and context — you might spend 18 months in a field site talking to dozens. Case study? In practice, could be one case. Could be four. The methodology drives the number, not the other way around Still holds up..
Why It Matters / Why People Care
Reviewers reject papers over this. Day to day, funders want a number in the budget. Practically speaking, ethics boards ask for justification. And if you say "I'll interview until saturation" without any concrete plan, you look unprepared.
But the real reason it matters? **Under-sampling misses complexity. Over-sampling wastes resources and participant goodwill.
I've seen studies with 40 interviews where the last 20 added nothing — just researcher exhaustion and participant burden. Which means i've also seen studies with 6 interviews claiming "saturation" on a topic that clearly needed 15. The findings were thin. The claims outran the data Easy to understand, harder to ignore..
There's also the credibility piece. Now, qualitative research already fights a perception battle. "Anecdotal.Also, " "Not generalizable. " "Small sample." When you can articulate why your sample size makes sense for your question, your methodology, your population — you defend the whole enterprise.
And let's be honest: participants give you their time, their stories, sometimes their trauma. Oversampling isn't just inefficient. It's ethically questionable Easy to understand, harder to ignore..
How to Determine Your Sample Size
This is where most guides give you a table. Still, " Those tables are useless without context. Ethnography: 30–50.Grounded theory: 20–60. In practice, "Phenomenology: 5–25. Here's how to actually think it through Most people skip this — try not to..
Start with your research question
Not your methodology. Your question.
"Are you exploring a shared essence?Consider this: " (Phenomenology — smaller, deeper) "Are you building theory from scratch? " (Grounded theory — larger, iterative) "Are you understanding culture in context?" (Ethnography — prolonged, many) "Are you evaluating a program or intervention?
The question dictates the methodology. Here's the thing — the methodology suggests a range. But your specific question narrows it further Which is the point..
Consider population heterogeneity
Homogeneous group — say, ICU nurses at one hospital who've all managed COVID deaths — you'll hit saturation faster. Maybe 8–12 interviews.
Heterogeneous group — ICU nurses across urban, rural, academic, community hospitals, different experience levels, different countries — you need more to capture variation. Maybe 20–30 That's the part that actually makes a difference..
This is maximum variation sampling in action. You're not sampling for representativeness. You're sampling for range.
Factor in data richness
A 90-minute life history interview yields more than a 20-minute semi-structured chat. A participant who's reflective, articulate, and emotionally engaged gives you more than someone who gives yes/no answers.
You can't always predict this. But you can design for it: longer interviews, multiple sessions, photo elicitation, timeline exercises. Richer data = fewer participants needed Simple, but easy to overlook..
Account for your analytic approach
Thematic analysis? You might need 20+ interactions to see patterns in talk. Even so, interpretive phenomenological analysis (IPA)? Idiographic focus means 3–6 cases, deeply analyzed. Now, discourse analysis? On the flip side, you can work with 10–15 solid interviews. Narrative analysis? Fewer cases, more time per case.
The analysis consumes the data. Know your appetite before you cook Easy to understand, harder to ignore..
Build in iteration
This is the part everyone forgets. Qualitative sampling is sequential, not simultaneous.
You don't recruit 20 people, interview them all, then analyze. Analyze. Analyze. Recruit 3 more. You recruit 5. Analyze again. Notice a gap — maybe no men, maybe no one under 30. Recruit purposively to fill it. Then assess saturation Practical, not theoretical..
This iterative loop is where rigor lives. "We interviewed 17 participants across three recruitment waves, with theoretical sampling guiding waves 2 and 3 after initial coding revealed...It's also where your sample size becomes defensible. " — that's a methods section reviewers respect Practical, not theoretical..
Practical benchmarks (with caveats)
Fine. You want numbers. Here are ranges I've seen work in practice, not as rules:
| Approach | Typical Range | When It Goes Higher |
|---|---|---|
| IPA / Phenomenology | 4–10 | Multiple perspectives on same phenomenon |
| Thematic Analysis | 10–20 | Heterogeneous population, multiple sites |
| Grounded Theory | 20–40+ | Building substantive theory, complex process |
| Ethnography | 20–50+ (informants) | Multi-sited, longitudinal |
| Case Study | 1–4 cases | Multiple embedded units per case |
| Focus Groups | 3–6 groups (5–8 each) | Need group dynamics, shared norms |
| Qualitative Evaluation | 15–30 | Multiple stakeholder groups, sites |
But — and I cannot stress this enough — **these are outcomes of good design, not inputs.But you pick it because your question demands theory generation. So naturally, ** You don't pick "grounded theory" because you want to interview 30 people. The 30 emerges.
Common Mistakes / What Most People Get Wrong
Common Mistakes / What Most People Get Wrong
Mistake #1: Confusing Convenience with Validity
People grab the first 15 participants they can reach and call it "saturation." But accessibility ≠ representativeness. A psychology student recruiting fellow students for a study on "help-seeking behavior among young adults" isn't studying young adults—it's studying undergraduates at one institution.
Theoretical sampling means going where you need to go, not where it's easy.
Mistake #2: Treating Qualitative Like Quantitative Quotas
"I need 30 participants, so I'll screen until I hit 30." This misses the point entirely. In qualitative work, you're not filling slots—you're building understanding. That said, if your early interviews reveal a crucial perspective missing from your sample, you don't just "find three more people. " You rethink your recruitment strategy And that's really what it comes down to..
Mistake #3: Stopping Too Early
Saturation isn't a fixed point you can measure with a stopwatch. And it's a judgment call based on emerging patterns. Some studies genuinely do saturate at 6 interviews. Others need 25. Some need 50 because the phenomenon is genuinely complex Nothing fancy..
The danger is stopping because you're tired, over budget, or your participant pool is drying up—not because analysis indicates saturation.
Mistake #4: Ignoring Analytic Capacity
You wouldn't drive 200 miles in a compact car, then complain about the suspension. If you're planning to do line-by-line coding, axial coding, memos, and theme development for each interview, factor in how long that actually takes. Same with qualitative data. Ten deeply analyzed interviews might exhaust your analytic bandwidth faster than 20 surface-level reads.
Mistake #5: Forgetting About Trustworthiness
Sample size arguments that ignore credibility, dependability, and confirmability aren't arguments at all—they're wishful thinking. Sometimes five interviews with intense reflexivity and audit trail beat 25 with poor documentation.
Making It Work for Your Project
Start with your research question. On the flip side, not your available participants. Not your IRB budget. Your question.
If you're exploring a well-understood phenomenon with established categories, you might need fewer interviews. Here's the thing — if you're unpacking something novel, you'll likely need more. If you're comparing groups, ensure each has adequate representation—not just total numbers Simple, but easy to overlook. Less friction, more output..
Use pilot interviews strategically. Your interview guide? Practically speaking, what are you learning about your approach? First three interviews aren't just data collection—they're your analysis rehearsal. Your coding framework?
Document your reasoning. "We conducted 12 interviews based on theoretical sampling that revealed sufficient variation in X, Y, and Z domains" carries more weight than "We did 12 interviews because that's what fit our budget."
Consider mixed methods thoughtfully. A qualitative subsample of 8–10 from a larger quantitative study can provide rich context for statistical findings without requiring massive qualitative investment Still holds up..
The Bottom Line
There's no magic number. There's only thoughtful design.
Good qualitative research isn't about hitting a predetermined sample size—it's about building sufficient data to answer your question with integrity. Sometimes that's five interviews. Sometimes it's fifty. Often it's somewhere in between, with justification rooted in analysis, not convenience Not complicated — just consistent. No workaround needed..
Your sample size should emerge from your methodological reasoning, not precede it. Let your data tell you when it's ready to be heard.
Final thought: The most compelling qualitative studies don't lead with sample size—they lead with insight. The sample size follows naturally from the depth of understanding achieved. That's not just good practice; it's honest scholarship.
