Qualitative Research: Methods, Process and Example

„How many interviews do I need?“ is the question almost every empirical dissertation starts with, and most guides answer it with „it depends“. Solid numbers on this have existed for twenty years.

Qualitative research is an approach that studies a small number of cases in depth in order to understand how people experience, interpret and explain something. It asks open questions instead of fixed response options, selects its cases by criteria rather than at random, and analyses language rather than numbers. By the end of this article you will know which method fits your question, how to justify your sample size, and when a different design would serve you better.

📌 The key points

• Qualitative research goes deep, not wide.
• Saturation appeared after twelve of sixty interviews (Guest 2006).
• Cases are selected by criteria, never by random sampling.
• Semi-structured interviews and focus groups are the most common methods.
• Ethics approval comes before your first recording, not after.

What is qualitative research?

Qualitative research is the umbrella term for every empirical method whose result is an interpretation rather than a measurement: it gathers text, images and sound, places that material in its context, and reconstructs the meaning the people studied attach to their own actions.

The UK Data Service puts the material itself at the centre of its definition: „Qualitative data is non-numeric information, such as in-depth interview transcripts, diaries“ (UK Data Service, Qualitative data). That single sentence already explains why the analysis looks so different from a survey: there is nothing to add up.

The contrast with quantitative research begins with the question itself. If you want to know what percentage of students watch the news in the evening, you need numbers. If you want to know why someone stopped and what that changed about their day, you need an account. Both questions are scientific, but they call for different tools.

Qualitative research does not tell you how many. It shows you how and why.

The limit of the approach follows from that same design. Because cases are chosen deliberately rather than at random, the findings apply to those cases first and not to a wider population. That is the design, not a flaw. Where the open, material-led way of thinking comes from is covered in the article on deductive and inductive research.

Qualitative or quantitative: what actually differs

The difference between qualitative and quantitative research is not the sample size but the question you are trying to answer. A study with twelve participants is not qualitative because twelve is a small number, but because it collects open answers and interprets them.

The belief that qualitative simply means „few cases“ is stubborn, and it costs marks every year. If you send out 200 questionnaires with five open text fields and then count how often each answer appears, you have worked quantitatively, however open the questions were. A study with eight carefully coded interviews, by contrast, remains qualitative.

The two terms themselves are explained in the article on qualitative vs quantitative. If you want the advantages and drawbacks of both approaches side by side, you will find them collected under qualitative and quantitative research methods.

Qualitative data collection methods

Qualitative data collection methods differ mainly in how close they get to the situation you want to learn about. An interview captures one person's interpretation. An observation sees the behaviour itself. A content analysis works with material that came into being without your involvement.

The choice therefore follows the research question rather than personal preference. If you are after motives and reasoning, there is almost no way around a conversation. If you are after routines the participants themselves barely notice, observation is more honest than any follow-up question. And if your subject already consists of texts or images, you do not need to interview anyone at all.

The semi-structured interview: the standard choice for dissertations

The semi-structured interview is by far the most common method in student work, because it produces rich material for a manageable amount of effort. You prepare a topic guide of open questions, then follow it loosely rather than slavishly, probing wherever something interesting appears. The guide is a handrail, not a script. How to word those questions is covered in detail under qualitative surveys.

Qualitative observation: open rather than pre-coded

Qualitative observation is frequently confused with its quantitative counterpart. The moment you set up a fixed tally sheet in advance and simply count instances in the field, you are collecting frequencies and therefore working quantitatively. Observation stays qualitative as long as you write detailed field notes instead of counting. Whether your categories come from theory or emerge from the material is a separate, independent decision.

⚠️ Caution

Running an observation with a ready-made tally sheet and calling it „qualitative“ in your methodology chapter misdescribes your own procedure. Decide before you enter the field whether the categories are fixed in advance or emerge afterwards, and write exactly that into the dissertation.

How many interviews do you need?

How many interviews you need cannot be derived from a formula, but there are empirical reference points: in the study by Guest, Bunce and Johnson (2006), saturation occurred within the first twelve of sixty interviews, and the basic elements of the overarching themes were present after as few as six.

Saturation here means the point at which further conversations stop producing new themes. Guest and his colleagues counted this systematically in their own material rather than simply asserting it, and their paper is still cited heavily on this question (Guest, Bunce and Johnson 2006: How Many Interviews Are Enough?, Field Methods 18(1), pp. 59 to 82).

Saturation: the documented ranges

A systematic review by Hennink and Kaiser (2022) analysed 23 papers on saturation, 17 of which used empirical data of their own. Across those 17 the range turned out to be remarkably narrow: saturation was reached at 9 to 17 interviews or 4 to 8 focus group discussions. The authors state a condition explicitly, and it matters more than the figure itself: this applies above all to relatively homogenous study populations and narrowly defined objectives (Social Science & Medicine 292, article 114523).

Two terms are easily confused here, and the difference belongs in every methodology chapter. The figures above measure when no new themes appear. Theoretical saturation in grounded theory means something else: the point at which no theoretically relevant differences are left to find. Justifying a sample size with the wrong concept is a common and avoidable slip.

How to justify your sample size

Two things follow for your own work. First, ten to fifteen interviews are a defensible range for a narrowly framed undergraduate or Master's project. Second, as soon as your group is mixed, you are comparing regions, backgrounds or age groups, or you are looking for fine differences in meaning, you will need more, and the figures above no longer carry your justification.

💡 Tip

Never put the sample size into your methodology chapter as a bare number. Write it as a decision with a reason: „Eight interviews were planned because all participants had been through the same programme. After the seventh conversation, no new themes emerged.“ Those two sentences make the number traceable; a bare figure stays open to challenge.

Your department's own guidance takes precedence in the end. Some programmes name their own minimum figures, and those override any recommendation from the literature.

Sampling in qualitative research

Sampling in qualitative research follows criteria rather than chance. Where a quantitative study draws at random so that the sample mirrors the population, you deliberately look for the cases in which your subject shows itself most clearly.

Calculating a sample size the quantitative way is therefore the wrong sum here. What you are after is range, not the average. Someone researching reasons for dropping out will deliberately seek both the first-year engineering student and the trainee teacher weeks away from finals. Only the contrast between the two shows what the cases have in common.

Two routes are common

• Predetermined selection: you set the criteria from your theory before entering the field.
• Theoretical sampling: you decide after each case who you need next.

With theoretical sampling the sample is not fixed in advance but develops as the study proceeds, and you stop once no theoretically relevant differences are left to find. For a dissertation with a fixed submission date the first route is usually the realistic one, because with the second you cannot say when you will be finished.

How a qualitative study is judged at the end is a separate question: objectivity, reliability and validity come from the quantitative tradition and do not transfer unchanged. Which frameworks apply instead is covered under quality criteria in qualitative and quantitative research.

Ethics approval before your first interview

Ethics approval has to be in place before you record your first interview, not once the transcripts are sitting on your laptop. Almost every UK university requires a Research Ethics Committee review for student projects involving human participants, and starting fieldwork without it can invalidate the data you have already gathered.

The framework most departments build on comes from the Economic and Social Research Council. Its core principles state that „wherever possible, participation should be voluntary and appropriately informed“, and the ESRC is explicit that ethics run across the whole project lifecycle, archiving and data linkage included (ESRC, Framework for research ethics).

What a committee usually wants to see

• A participant information sheet in plain English.
• A consent form covering recording, storage and later use.
• Your plan for anonymising transcripts and quotations.
• Where the recordings will live and when they will be deleted.
• What happens if a participant withdraws after the interview.

Build the review time into your schedule rather than treating it as paperwork at the end. Committees meet on fixed dates, and missing one can cost several weeks in a project that only has a few months to begin with.

The terminology around the written work itself varies more than most students expect. Oxford's guidance states that „a dissertation is the final piece of writing submitted for a Master's degree“ and notes that the two words are used the other way round in the United States (University of Oxford, Writing Theses and Dissertations). Check your own department's wording rather than assuming. At doctoral level the oral examination is standard: Sheffield describes it as „a regulatory requirement for all students undertaking a higher degree“.

The process of a qualitative study

The process of a qualitative study runs from question to finding in six steps, and unlike a survey you are allowed to go back. If you notice at step four that the topic guide is missing the point, you revise it and conduct the next conversations differently.

That flexibility is not a licence to improvise silently. Every change to the procedure has to be documented, otherwise nobody can follow the study afterwards. What you describe in the methodology chapter must match what actually happened.

1. Sharpen the question. „How do students experience the news?“ becomes „How do third-year education students explain having stopped watching the news in the evening?“. Only that narrowing makes case selection decidable. More on this under criteria for good research questions and hypotheses.
2. Choose the method and justify it. Interview, observation or content analysis follows from the question. Write the reasoning down in a paragraph; it belongs word for word in the methodology chapter later.
3. Select cases and secure access. Who do you need, how many, and through which route do you reach them? Access to the field causes most delays in practice.
4. Collect and record. Test the recorder, obtain written consent, arrange a quiet room. An unusable recording costs you the whole conversation.
5. Transcribe and code. Sound becomes text, text becomes categories. Budget generously for this, as typing up takes several times the length of the conversation.
6. Analyse and situate. Categories become statements, and statements are set against the literature.

For the analysis itself, thematic analysis and qualitative content analysis are the two routes most commonly taught. Whichever you choose, name it in the methodology chapter and follow its steps. How data collection fits into the wider research process is covered in the article on data collection.

When qualitative research is the wrong choice

Qualitative research is the wrong choice as soon as your question asks for a quantity, a proportion or a tested relationship. Twelve interviews tell you which explanations exist, not how widespread they are, and that confusion produces a findings chapter any second marker can pull apart.

In these six cases you need a different design:

• Your question begins with „how many“, „how often“ or „how strongly“.
• You want to make a claim about a whole group rather than your cases.
• You want to compare two time points or two groups statistically.
• Your supervisor has explicitly asked for a significance test.
• You have no access to people willing to talk to you at length.
• You have fewer than eight weeks left for fieldwork and analysis.

The time budget is the point most often underestimated. Between the first scheduling email and finished coding, twelve interviews take several weeks of pure working time, and most of that is transcription.

The mistake about cause and effect

You will often read that qualitative research uncovers cause and effect. That holds only in a very narrow sense. An interview tells you which connection a person themselves draws between two things, which is a valuable finding in its own right. Whether that connection holds beyond the individual case has not been tested. A tested claim about effects requires a hypothesis-testing design and a sufficiently large, randomly drawn sample.

Mixed methods: both routes together

Qualitative and quantitative components can be combined, a design known as mixed methods. Three basic forms are common: both strands in parallel, qualitative first and quantitative second, or quantitative first with a qualitative strand to explain the findings. The deciding question is which of your questions cannot be answered by one approach alone. One honest warning belongs here: in practice the two strands are rarely brought together properly, and two rounds of data collection almost always mean substantially more work.

Qualitative research example

A qualitative research example makes the steps easier to picture than any definition. Suppose you are writing your dissertation on news avoidance and want to know why young people stop following the news.

The example is invented, the procedure is not: this is how an interview study in a dissertation actually runs, from the first narrowing of the question to the sentence that ends up in the limitations.

The research question

„How do students aged 20 to 26 explain having deliberately cut down their news consumption?“ The question is narrow enough for a dissertation and open enough to invite accounts. You do not need a hypothesis here, because you are exploring rather than testing.

Choosing the method

You settle on the semi-structured interview, because you need reasoning and interpretation, and both only appear in conversation. Observation is out, since news avoidance consists of not doing something. A diary study would be possible but demands weeks of commitment and would answer the question only indirectly.

The sample size

You plan twelve interviews. The justification: the group is homogenous in age and life situation, the question is narrowly framed, and for exactly this constellation the review by Hennink and Kaiser (2022) reports a range of 9 to 17 interviews. After the eleventh conversation no new patterns appear, and the twelfth confirms the finding.

The analysis

You transcribe every conversation, read each twice, and assign categories close to the wording: overload, the same bad news over and over, powerlessness, a deliberate time limit. You then summarise how the categories relate to each other and set the result against existing research.

What you can claim, and what you cannot

Your finding reads: there are four recurring patterns of explanation for news avoidance, and three of them describe exhaustion rather than indifference. What you cannot say is what percentage of students in the UK avoid the news. That sentence belongs in your limitations, and it makes the dissertation stronger rather than weaker.

Conclusion

Qualitative research is worth choosing whenever you want to understand how people experience and explain something. Decide early and write the reasoning down, because supervisors reward a decision they can follow more highly than a method that merely looks demanding. A small, well-justified interview study beats a large survey without a concept.

Where to go next

• Want to build your topic guide? How to create a questionnaire
• Need the overview of every phase? The empirical research process
• Want to know how qualitative surveys are analysed? Qualitative surveys: forms and analysis
• Looking for the foundations? Empirical research: definition and guide

Collecting open answers without a pile of paper?

With empirio.ai, an online survey tool from Germany, you can ask open questions, gather the answers in one place and export them for analysis.

Create a survey for free