Qualitative Data Analysis Methods

Choosing a qualitative analysis method is where a lot of methodology chapters quietly go wrong. In our coaching sessions, we often meet students who picked thematic analysis because everyone else did, or grounded theory because it sounded rigorous, and then described one method while doing another. Your committee (your examiners, in the UK) will notice, because naming a method is a promise about how you handled your data.

In our work with 10,000+ students, the fix often starts with one question: what is your research question actually asking? In this post, I’ll explain the seven methods you’re most likely to meet, compare them side by side, give you a rule of thumb for matching your question to a method, and show you how five real dissertations named and applied theirs.

---

What is qualitative data analysis?

Qualitative data analysis is the process of organizing and interpreting non-numerical data, such as interview transcripts, documents or field notes, so that it answers your research question. Where quantitative analysis runs statistics on numbers, qualitative analysis works with words, images and observations, and the findings are usually themes, categories, accounts or a theory rather than figures.

Qualitative data covers a lot more than interviews. Here are ten common kinds you’ll meet in a dissertation, grouped by where they come from:

If you’re weighing up which kind of research to do in the first place, our guide to qualitative vs quantitative research covers the differences. The short version is a trade-off between depth and breadth:

“So, is qualitative analysis the easier option?”

Not really. It takes no statistics, but it takes a lot of time. By the end of data collection you can easily have hundreds of pages of transcripts, plus field notes and the impressions you never wrote down. Turning all of that into findings you can defend means reading and re-reading, coding, comparing and rewriting, and in our experience it usually takes longer than students plan for. Budget for it.

---

The seven main qualitative analysis methods

There are dozens of named approaches to qualitative analysis, but seven come up again and again in the dissertations and theses we coach:

1. Content analysis
2. Narrative analysis
3. Discourse analysis
4. Thematic analysis
5. Grounded theory
6. Interpretative phenomenological analysis (IPA)
7. Framework analysis

They overlap a good deal, because most of them involve coding your data at some point. What separates them is what aspect of the data they focus in on, and what you end up with:

Method	What it looks at	What you end up with
Content analysis	What texts contain, coded into categories and often counted	Categories and patterns across many texts
Narrative analysis	The stories people tell, and how they tell them	An account of how people make sense of events
Discourse analysis	Language in its social context, including power	An account of how language shapes a social reality
Thematic analysis	Patterns of meaning across a dataset	A set of themes, each backed by extracts
Grounded theory	A process with little existing theory	A new theory, built from data collected in rounds
IPA	How a few people make sense of a major experience	Detailed case-by-case accounts, then shared themes
Framework analysis	The same topics across every case, compared in a matrix	A chart of cases by codes, then themes across it

Keep that last column in mind as you read on. If you know what your findings chapter needs to deliver, you’re most of the way to choosing your method.

Qualitative content analysis

Content analysis is one of the most common and most straightforward methods. You code the content of texts (or images, or video) into categories, then look for patterns within or across them. It works well on large collections: newspaper coverage, policy documents, social media posts or speeches.

It can stay close to the surface or dig deeper. At the surface, you might count how often a concept appears, such as how frequently news reports describe nurses as heroes. Deeper down, you might look at what the wording implies, such as how tourism brochures present a country as ancient or modern. Our content analysis guide calls this the difference between explicit and implicit content.

Because content analysis often counts, it brings a small dose of quantitative thinking into a qualitative study. That’s its strength, and also its main criticism: counting can strip a phrase of the context that gave it meaning. It also needs a tightly defined question, or you’ll code everything and conclude nothing.

A political science PhD from Cornell gives a clean example. It analyzed more than 500 presidential commemorative speeches from South Korea and Taiwan, given between 1948 and 2022. Watch for the rule it sets for borderline cases:

Notice how the passage names two things to code, gives each a three-point scale, defines them (we’ve cut the second definition for length), and then decides in advance what happens to a speech that doesn’t clearly fit. A footnote on the same page adds that the coding was cross-checked with eight research assistants. If you use content analysis, write your own decision rules into your codebook before you start, so two readers would code the same speech the same way.

Our content analysis video covers the what, why and how of the method, with examples:

Narrative analysis

Narrative analysis is about the stories people tell, and how they tell them. Stories are how people make sense of what has happened to them, so analyzing their structure, sequence and emphasis can show how someone understands their own experience.

You might analyze how entrepreneurs describe the setbacks in their careers, or how patients with a chronic illness tell the story of their diagnosis. The question isn’t only what happened. It’s which events the teller puts first, who they cast as responsible, and where the turning point falls.

Narrative studies usually work with a small number of participants, because each story is long and read closely. That limits how far you can generalize, and it puts your own interpretation front and center, so you’ll need to be open about how your background shapes what you see. Our guide to research bias covers the main traps.

Next, a Purdue PhD in English that studied community organizers in Appalachia, using participants’ photos and written stories, focus groups and interviews. Look at the order in which the stories were analyzed:

The analysis runs in two passes: first the familiar stories told about the region, then the stories that push back against them. For the second pass it borrows grounded theory’s coding procedure, names it and cites it, without claiming to be a grounded theory study. That’s fine, as long as you do what this passage does: say which method did which job. Your findings chapters can then follow the shape of the analysis, as this thesis’s chapters do.

For a simple explainer on narrative analysis, with worked examples, watch this:

Discourse analysis

Discourse is a fancy word for language in use, written or spoken. Discourse analysis looks at language within its social context: how the way people talk about something reflects, and helps create, the culture and power relations around it. You could analyze how managers and junior staff speak to each other in a company, or how politicians talk about immigration after an election.

Context does a lot of the work here. A junior employee might speak casually to the chief executive in a company that prides itself on a flat structure, and very formally in one that doesn’t. Discourse analysis asks what that difference tells you about the organization.

Approaches range from close readings of grammar and word choice to critical discourse analysis, which focuses on power. If your research aims involve culture, identity or power, discourse analysis is a strong candidate. Because almost any text can be read this way, you need a precise research question, or the analysis has no edges.

Discourse analysis is easy to confuse with content analysis, since both often work with documents. The political science thesis quoted above names the difference well: it describes its hand-coded content analysis as trying to “bridge the interpretive foundations of discourse analysis and narrative comparability afforded by content analysis” (Jo, 2024, p. 56). In other words, discourse analysis interprets what language does, and content analysis makes texts comparable by coding what they contain. Decide which of those your question needs.

If discourse analysis is on your shortlist, this video explains how it works:

Thematic analysis

Thematic analysis identifies patterns of meaning, called themes, across a dataset, such as a set of interview or focus group transcripts. You code the data, group related codes, and develop themes that answer your research question. Virginia Braun and Victoria Clarke’s six-phase approach is the best-known version, and it’s the one the thesis below cites.

Thematic analysis is popular because it’s flexible. It isn’t tied to one theory, so it suits a wide range of questions about people’s experiences, views and opinions. The catch is that flexibility needs explaining: your committee will want to know which version you used and why.

Here’s how an education master’s thesis from New Zealand handles that. It studied teachers’ perspectives on their professional standards (EC, below, stands for early childhood). Notice how it explains its version before listing any steps:

The thesis names a theoretical (deductive) approach rather than an inductive one, ties that choice to its framework, and admits the framework shaped what was noticed during coding. That admission helps the thesis: your committee can see how the framework steered the coding. Do the same in yours: say whether your themes came from the data or from a framework, and why. Our post on inductive and deductive coding explains the difference.

Two practical warnings. Thematic analysis often shifts as you go, and your themes (and in inductive work, sometimes even your research question) can change part-way, so budget extra time for rework. And a list of themes isn’t the finish line: your findings need to say what the themes mean together.

To see codes turn into themes, watch our thematic analysis explainer:

Grounded theory

Grounded theory is used to build a new theory from your data, rather than test an existing one. Strictly speaking it’s a whole research design rather than just an analysis method, because it shapes how you sample and collect data as well as how you analyze it. It’s included here because students so often meet it as an analysis choice.

What makes it grounded theory is a set of practices: constant comparison of new data with what you’ve already coded, memo-writing, and theoretical sampling. Instead of choosing a representative sample up front, you let the analysis decide who you talk to next. As one widely cited design guide puts it, theoretical sampling “is used to identify and follow clues from the analysis, fill gaps, clarify uncertainties, check hunches and test interpretations as the study progresses.” You keep going until your categories stop developing.

Here’s how those practices fit together as a loop, with saturation as the way out:

Grounded theory suits topics where little existing theory explains what’s going on. It’s demanding, though. You need to collect data in rounds, which can be hard to fit into a one-year master’s. And the different schools disagree about the literature: classic grounded theory says to stay away from the literature in your area until your analysis is done, while some later versions, such as Strauss and Corbin’s, let you use it earlier. Pick a school and follow it.

With grounded theory, the usual trouble is a study that calls itself grounded theory but delivers a list of themes. If your study claims to be grounded theory but isn’t sampling theoretically or building a theory, it’s probably thematic analysis, so call it that. Our full guide to grounded theory traces two real PhD analyses step by step.

Interpretative phenomenological analysis (IPA)

Interpretative phenomenological analysis, or IPA, explores how individual people make sense of a major experience: a diagnosis, becoming a parent, losing a job, living through a disaster. The experience is the “phenomenon” in the name. IPA suits questions about lived experience, asked of a small number of people who share that experience.

The “interpretative” part matters. As a short paper on IPA by Jonathan Smith and Mike Osborn puts it, “the researcher is trying to make sense of the participant trying to make sense of what is happening to them.” IPA is also idiographic: it examines each case in detail before moving to general claims, which is why samples are usually small.

That focus on the individual is easy to lose. You’ll still code your data, but if you reduce every account to a handful of codes, you lose the depth that justified IPA in the first place. And because your interpretation is part of the analysis, you need to be explicit about how your own experience shapes it.

IPA is only one kind of phenomenological analysis, and the difference is worth seeing. A nursing PhD from the University of Massachusetts Amherst shows the contrast. It studied nurses who had witnessed dying patients’ end-of-life dreams and visions (ELDVs). It used descriptive phenomenology, an older approach, instead of IPA. Look at what it asks the researcher to do with their own experience:

Descriptive phenomenology tries to set the researcher’s experience aside. IPA treats the researcher’s interpretation as part of the analysis, and asks you to make it visible instead. Neither is wrong, but they aren’t interchangeable, so check which one your design really follows. Then name it precisely, along with the analysis procedure that goes with it (this thesis names Colaizzi’s method a page earlier).

Framework analysis

Framework analysis, sometimes called the framework method, is most at home in health and policy research. Gale and colleagues’ guide to the method places it within “a broad family of analysis methods often termed thematic analysis or qualitative content analysis”. Its defining feature is a matrix of cases (rows) by codes (columns), which lets you compare across participants and across codes at the same time.

The same guide recommends it for interview data where you want to compare within and between cases, and notes that it works well in research teams, even where not every member has done qualitative research before. Its limit is the flip side of the matrix: it can’t handle highly heterogeneous data, so your interviews need to cover similar topics or issues. Individual participants can still say very different things about each one.

A public health PhD from the University of Calgary (in community health sciences) used it on interviews with 13 international experts in health technology reassessment and knowledge translation. Look at how the passage justifies the choice before listing any steps:

The passage gives a reason for the choice, credits the method’s originators and lists its steps. Note that authors count those steps differently: the guide linked above breaks the same work into seven stages. Cite the version you follow, and keep its step names consistent through your chapter.

How to choose the right analysis method

Like every other methodological decision, your choice of analysis method follows from your research aims, objectives and questions. The wording of your research question is usually the best clue:

• “How often…?” or “How is X represented in these documents?” points to content analysis.
• “What are people’s views or experiences of X?” points to thematic analysis.
• “How do people tell the story of X?” points to narrative analysis.
• “How is X talked about, and with what effect?” points to discourse analysis.
• “What process explains how X happens?”, where little theory exists, points to grounded theory.
• “How do these people make sense of X?”, asked of a few people, points to IPA.
• “How do these cases compare on the same set of issues?” points to framework analysis.

These are leanings, not rules, but if your question and your method point in different directions, your committee will spot it. Our guide to choosing your analysis method works through content, thematic, narrative and discourse analysis as a decision.

Once you have a leaning, check it against three practical things. First, your data: IPA needs rich accounts from a few people, while content analysis copes with hundreds of documents. Second, your research philosophy, since some methods carry assumptions about knowledge that others don’t. Third, your time, because grounded theory’s rounds of data collection need months a short program may not have.

“Can I use more than one method?”

Yes, as the Appalachia thesis shows. Combining methods is one form of triangulation, and it can give you a fuller picture. What matters is that each method has a clear job. A second method you can’t explain only adds risk.

The mistake I see most often isn’t choosing the wrong method. It’s naming a method without its tradition: calling an analysis IPA without the case-by-case reading, for example. Never pick a method because it sounds impressive or because you used it before. Name the method and its version precisely, cite the source you followed, and show each step with a short example from your data.

Once your method is set, the next job is writing it up. Our guide to writing the methodology chapter shows how real dissertations justify each choice, and our qualitative findings guide covers how to present what your analysis found.

A method you can name, cite and show step by step is one your committee can’t easily pick apart.

Still have questions?

What are the 5 methods of qualitative data analysis?

There’s no single official list, but five methods appear on many lists: content, thematic, narrative and discourse analysis, and grounded theory. Fuller lists often add interpretative phenomenological analysis (IPA) and framework analysis, which are widely used in psychology and health research. The right one depends on your research question, not on which list you read.

What are 10 examples of qualitative data?

Common examples are interview transcripts, focus group recordings, open-ended survey answers, diaries or journals, field notes from observation, photos or video, documents and policies, social media posts, news articles, and speeches. All of them record words, images or behavior rather than numbers.

Which qualitative analysis method is the most common?

In the dissertations we coach, thematic analysis is the method we see most, because it’s flexible, isn’t tied to one theory and suits questions about people’s views and experiences. Content analysis is also very common, especially for documents and media. Popularity isn’t a reason to choose either: the method has to fit your research question.

Thematic vs content analysis: what’s the difference?

Both code qualitative data, but they aim at different things. Content analysis categorizes what texts contain and often counts how frequently categories appear, which suits large collections of documents. Thematic analysis develops themes that capture patterns of meaning across a dataset, which suits questions about people’s experiences and views. If your findings will include frequencies, content analysis is usually the better fit.

Can you use more than one qualitative analysis method?

Yes, as long as each method has a clear purpose. A study might use narrative analysis to identify the stories participants tell and thematic analysis to compare those stories across participants. Say in your methodology chapter which method did which job, and cite a source for each, so your committee or examiners can follow the logic.

Can’t find your answer here? Ask a Grad Coach directly – the initial chat is free.

Speak with a friendly coach →

Cite-worthy sources

Cite this article

Warren, K. (2020). Qualitative Data Analysis Methods. Grad Coach. https://gradcoach.com/qualitative-data-analysis-methods/

Copy citationMore citation formats

• Barnes, P. D. (2019). Teacher standards and professionalism [Master’s thesis, Victoria University of Wellington].
• Carlson, E. B. (2019). “There is wealth in the struggle”: Unearthing and embracing community knowledges through organizing work in Appalachia [Doctoral dissertation, Purdue University].
• Chun Tie, Y., Birks, M., & Francis, K. (2019). Grounded theory research: A design framework for novice researchers. SAGE Open Medicine, 7, 2050312118822927.
• Esmail, R. (2020). Understanding the relationship between health technology reassessment and knowledge translation [Doctoral dissertation, University of Calgary].
• Gale, N. K., Heath, G., Cameron, E., Rashid, S., & Redwood, S. (2013). Using the framework method for the analysis of qualitative data in multi-disciplinary health research. BMC Medical Research Methodology, 13, 117.
• Herron, S. L. (2025). The lived experience of nurses witnessing end-of-life dreams and visions of dying patients [Doctoral dissertation, University of Massachusetts Amherst].
• Jo, E. A. (2024). Nations after democracy: Politics of storytelling in South Korea and Taiwan [Doctoral dissertation, Cornell University].
• Smith, J. A., & Osborn, M. (2015). Interpretative phenomenological analysis as a useful methodology for research on the lived experience of pain. British Journal of Pain, 9(1), 41-42.