Grounded Theory:
What It Is and How It Works

At first encounter, grounded theory can sound like the open-minded choice: no framework, no hypotheses, just let the data speak. So plenty of students pick it, write “a grounded theory approach was used” in their methodology chapter, and do something else entirely. Then an examiner asks them to walk through how they got from their data to their theory, and there’s no theory to walk to.

In our 10 years of helping students with their methodology, grounded theory is one of the research designs we most often see students claim and then struggle to defend. In this post, I’ll explain what grounded theory actually commits you to, how its three schools differ, and how to tell it apart from thematic analysis. Then I’ll trace two real doctoral analyses from a participant’s words toward a theory, including where an examiner would push.

---

What is grounded theory?

Grounded theory is a qualitative research design in which you develop a theory from your data, instead of starting with a theory and testing it. Barney Glaser and Anselm Strauss set it out in 1967, in The Discovery of Grounded Theory, drawing on their study of how hospital staff handled dying patients. They put the idea in one line:

The basic theme in our book is the discovery of theory from data systematically obtained from social research.

The book pushed back against research that mainly set out to verify existing theories. Glaser and Strauss wanted researchers to generate theory from the data up, which is where the name comes from. That’s the part everyone remembers.

What gets forgotten is the machinery. Grounded theory isn’t a mood or an attitude toward your data. It’s four specific practices, and a study that skips them isn’t grounded theory, regardless of what its methodology chapter says:

1. Constant comparison. You compare each new piece of data with the data you’ve already coded, then codes with codes, then categories with categories. Similarities and differences are how categories form.
2. Theoretical sampling. You decide who to talk to next based on what your analysis needs. Glaser and Strauss define it as the process “whereby the analyst jointly collects, codes, and analyzes his data and decides what data to collect next and where to find them, in order to develop his theory as it emerges.”
3. Memo-writing. You write notes to yourself about your codes and categories as you go. These memos are where the theory actually gets built.
4. Saturation. You stop sampling when your categories stop developing.

That last one is widely misunderstood, so it’s worth reading the original.

Saturation doesn’t mean “my last few interviews didn’t raise anything new.” Glaser and Strauss tie it to a category: “Saturation means that no additional data are being found whereby the sociologist can develop properties of the category.” In other words, you’ve saturated a category when new data stops adding to what you know about it.

Kathy Charmaz, whose version I’ll come to below, accepts that definition, and cites another qualitative methodologist’s warning that researchers often proclaim saturation rather than prove it, so treat saturation as something you argue for, not a box you tick.

Notice what the four practices have in common: they all depend on collecting and analyzing at the same time. It’s hard to decide what to collect next if you’ve already collected everything. That one fact makes grounded theory a poor fit for a lot of dissertation designs, as you’ll see later.

Here’s how they fit together. Three of the practices form a loop with data collection. The fourth, saturation, isn’t part of the loop at all: it’s how you exit the loop.

The three schools of grounded theory

Glaser and Strauss later disagreed about how the method should work, and a third version later emerged as a critique of both. In 1992, Glaser published a book attacking the version Strauss had developed with Juliet Corbin. As one journal article on the split summarizes it, his charge was that by using concepts like “axial coding” and “coding paradigms,” researchers would “force” categories on the data instead of allowing them to “emerge.”

Kathy Charmaz, who had studied with both of them, then built a third version that rejects the idea of “discovering” a theory at all.

So when you say you’re using grounded theory, your examiners will want to know whose. Here’s how the three compare:

The classic and Straussian coding labels follow a published comparison of the three schools; Charmaz’s are from her own book. The difference that matters most in practice is the first row. Charmaz puts her position bluntly in Constructing Grounded Theory:

Unlike their position, I assume that neither data nor theories are discovered. Rather, we are part of the world we study and the data we collect.

It’s the version both examples below use, and it suits researchers who already reflect on their own influence on their data. It doesn’t ask you to pretend you arrived with an empty head.

“So which one should I pick?”

Pick the one whose assumptions match your research philosophy, then follow it consistently. If you’re an interpretivist who reflects on your own influence, Charmaz is the natural fit. If you need a detailed procedure and your committee wants one, Strauss and Corbin give you the most structure.

Whatever you choose, cite that school’s own texts for its procedures and use its labels consistently. Calling the same step “open coding” in one paragraph and “initial coding” in the next, without saying why, is the kind of inconsistency we see examiners pick up on.

---

Grounded theory or thematic analysis?

This is the question I’d settle before anything else, because a lot of studies described as grounded theory are really thematic analysis. Thematic analysis is a perfectly good method. The trouble only starts when you claim one and deliver the other.

Virginia Braun and Victoria Clarke, authors of the best-known guide to thematic analysis, describe this pattern directly in their paper on the method:

However, in our experience, grounded theory seems increasingly to be used in a way that is essentially grounded theory ‘lite’ - as a set of procedures for coding data very much akin to thematic analysis.

Their point is that a “full-fat” grounded theory “requires analysis to be directed towards theory development.” If you’re not developing a theory, they argue, naming your method as thematic analysis means you needn’t take on grounded theory’s commitments at all. So the question to ask of your own study is whether it’s building a theory.

The table below shows where the two methods part ways in practice:

	Grounded theory	Thematic analysis
Aim	Build a theory that explains a process or problem	Describe and interpret patterns of meaning
Data collection	In rounds, alongside analysis	Can all be collected first, then analyzed
Sampling	Theoretical: the analysis decides who’s next	Can be decided up front
When you stop	When your categories are saturated	When you’ve analyzed the dataset you planned
What you end with	Categories integrated into one theory, often around a core category	A set of themes
The literature	Depends on the school, often delayed	Can come first, especially in deductive work

A quick test: look at your planned findings chapter. If it reads as “Theme 1, Theme 2, Theme 3, Theme 4” with no account of how they connect into a process, it’s probably closer to thematic analysis than grounded theory.

A recent methods note in the Grounded Theory Review, written from the classic Glaserian tradition, makes the same point and lists sentences that give a study away, including “The five themes describe participants’ experiences.” As it puts it, “Themes are neither a stage in grounded theory analysis nor an intermediate step on the way to theory.”

So if you’re coding interviews inductively but you’re not sampling theoretically or building a theory, call it thematic analysis, or inductive coding within it. You’ll have a cleaner chapter and a much easier viva.

If thematic analysis sounds like the better fit, this video explains how it works, with examples.

Two real grounded theory examples

That’s probably all rather conceptual, so let’s look at some real-world examples. Here’s what grounded theory looks like on the page, in two successful, published doctoral theses. Both used Charmaz’s constructivist version, and both show their working in their methodology chapters, which makes them worth studying. I’ll walk through one thread of each, then show where an examiner might push.

The figure below sets the two threads side by side, from a participant’s words up to the top of each analysis. Notice where each one ends.

Example 1: place enters a theory of identity

The first is a psychology PhD from the University of Hawaiʻi at Mānoa, which asked how intersectionality (“race, gender, sexuality, and class”) shapes the development of a multiracial identity. Stevy Scarbrough analyzed free-drawing and writing exercises and semi-structured interviews with 23 participants, using Charmaz’s 2014 edition.

Her first participant drew the ocean and a palm tree, and explained why. Here’s the open coding of that explanation, as she sets it out in her Table 2:

Line	What the participant said	Open codes
19	But throughout the middle, I drew the ocean	Place represents identity
20	and a palm tree to represent ______	Ocean and palm tree symbol of living on an island
21	because I was born and raised here, and I feel
22	like that had a lot to do with who I am.	Location raised important to identity

The blank on line 20 is in the original. Notice how close the codes stay to the data: they’re short, specific and still in the participant’s world. Scarbrough underlined “Place represents identity” as one of her overarching codes, and place appears again among the 20 concepts from her initial coding of the interviews, as “Place informs identity.”

She also wrote a memo on the drawing. It’s worth reading closely, because it does something memos are supposed to do: it catches a mistake.

Two things to take from it. First, the memo corrects her own earlier code: in her open coding of the drawing, she’d read one shape as “A hut, place been to?” The memo strikes through “and a hut” and notes that it was actually a food item, musubi. Memos are where you catch yourself, so write them while your codes are still provisional.

Second, it puts her interpretation and the participant’s own account side by side, which is constructivist grounded theory doing exactly what it says.

Through focused coding, place became one of her five categories, which she labels as themes: “Theme 4. Place is Salient.”

It also became the outer ring of her final model, in which the individual sits inside their family, and the family sits inside “a place-based context.” And here’s the most grounded part. Place wasn’t in her research question at all. In her discussion, she notes that place “was found to be an important aspect of multiracial identity, in addition to the four social identities that made up my research question.”

That’s the payoff of grounded theory in one sentence: the data told her something her question didn’t ask, and her method was open enough to let it in. If your analysis only ever finds what your research question listed, it’s worth asking whether you were really building theory or confirming a plan.

Example 2: reaching a core category

The second is a nursing PhD from Queen’s University in Canada, on how people with diabetic foot ulcers come to engage in managing their own care. Idevania Costa interviewed 30 people at a wound care clinic in Ontario, using theoretical sampling and follow-up calls.

Costa coded “idea-by-idea” rather than line-by-line, because the participants’ answers were long and full. Where possible, Costa used in vivo codes, which are participants’ own words. Several participants described their own self-care in similar terms:

Those lines became the in vivo code “being a bad diabetic.” Then constant comparison took over. Costa went back through all the data to see how each participant talked about looking after their diabetes, and what sat around the code: what was associated with it, how it related to their first ulcer, and what support they had. That comparison turned a label into a concept, which went into a category called “Perception of Illness.”

Here’s where Costa’s analysis goes further than Scarbrough’s.

“Perception of Illness” became Phase 1 of five phases in a single process, which the thesis names as its core category: “The Journey Toward Engagement in SCM” (self-care management). The other phases run from noticing foot changes, through realizing how serious it was, to learning about and engaging in self-care, with contextual factors like aging and support shaping the whole process.

Everything in the analysis hangs off that one process. That’s what a core category looks like: not the biggest theme, but the account that ties all the others together.

Costa’s thesis also shows theoretical sampling in practice. After the first five interviews, Costa and their supervisor revised the research questions and interview guide, and later participants were recruited “according to emergent categories.” The categories also “became the foundation of the follow-up interview questions.” That’s the method’s engine running: the analysis deciding what to collect next.

One caveat if you go to read it: the public copy of the thesis stops partway through Chapter 4, so you can follow the method in full but not all of the findings. The version submitted for examination may have been more complete, but we don’t know.

What an examiner would ask

Both of these theses passed, and both are good models. They also both leave openings that an examiner could probe, and spotting them will help you defend your own.

Scarbrough interviewed each participant once, over a single ten-week period, from one university’s student research pool. She describes theoretical sampling as probing new ideas within interviews and using later interviews to fill gaps, and she claims saturation.

An examiner might ask how theoretical sampling decided who came next, when participants signed up for time slots from one pool. And she reports her findings as five themes and a model rather than one core category. That’s close to the pattern the Grounded Theory Review note warns about, so a Glaserian examiner would likely push on it; a constructivist one might accept the model as her theory, if she argued for it.

Costa’s analysis reaches a core category and shows theoretical sampling changing the interview questions. Everyone was also recruited through one clinic, so an examiner could ask a version of the same sampling question.

Neither is a criticism of the authors. These are the kinds of questions your examiners could ask of your study, so answer them in your methodology chapter before anyone else raises them: how your sampling followed your analysis, what saturation meant for each category, and how your categories connect into one theory.

Those three aren’t the only ones. Below are the seven areas we see examiners probe most often in grounded theory studies, each with the kind of question it tends to raise. Read them as a checklist for your own methodology chapter.

The coding stages in each school

Each school names its coding stages differently, and the names aren’t interchangeable. Here’s the short version, with links to our posts on each technique:

• Charmaz: initial coding, then focused coding. Charmaz describes “at least two main phases,” and treats axial and theoretical coding as optional extras. Following Glaser, she recommends coding with gerunds (“-ing” words like defining or avoiding), which keeps your codes on actions and processes rather than topics. That’s the idea behind process coding.
• Strauss and Corbin: open coding, then axial coding, which relates categories to subcategories using their “paradigm” (conditions, actions and consequences), then selective coding, which integrates everything around one core category.
• Glaser: open coding and selective coding (together, “substantive” coding), then theoretical coding, which works out how your categories relate to each other as a theory.

Two practical points. First, all three are iterative: you’ll move back and forth between stages, and that’s the method working. Second, use your chosen school’s labels consistently, because the vocabulary tells your examiner which procedure you claim to have followed. If you borrow a term from another school, say so, and say why.

---

Grounded theory in a dissertation

Grounded theory was first worked out in a grant-funded study, by researchers with the time to keep going back to the field. A dissertation usually looks quite different. Your university probably expects an ethics application and a literature review chapter before you’ve done a single interview. You’re working alone, to a fixed deadline that leaves little room for repeated rounds of data collection. And at the end, an examiner will want to see how you got from your data to your theory.

None of that rules grounded theory out. But each of these pulls against the way the method is meant to work, so it pays to plan for them from the start. Here’s how I’d handle each one, in that order.

Ethics approval for an evolving sample

Theoretical sampling means you can’t list all your participants and questions in advance, and ethics committees like to see both. A team running a grounded theory study of dental practices in Australia handled this by explaining in their application that grounded theory procedures “evolve,” providing “a long list of possible recruitment strategies and interview questions,” and stating that they would apply to modify their protocols as the study progressed. They filed several modifications, and each was approved “without difficulty.”

That’s one committee’s experience, not a guarantee. But the approach is sound: describe the evolving design up front, give your committee a broad list of possible questions and recruitment routes, and plan for amendments in your timeline. Our guide to IRB approval covers the process in more detail.

For the basics any ethics application has to cover, this video explains the four core considerations your committee will look for.

When to write the literature review

The classic advice is to stay away from the literature in your area until your analysis is done, so it doesn’t shape what you see. Glaser and Strauss recommended literally ignoring it at first. The trouble, as one journal article on the literature review in grounded theory points out, is that for PhD students “research funding, ethical approval and progression through the doctoral process may all be heavily dependent upon producing a detailed literature review prior to commencing primary data collection.”

The practical answer is the one Charmaz gives: do the review your program requires, then let it “lie fallow” until you’ve developed your categories, and write the full literature review after your analysis. Robert Thornberg goes a step further. What he calls “informed grounded theory” uses the literature throughout the study, as a source of ideas rather than a frame you force onto your data, because, as another methodologist put it, there’s “a difference between an open mind and empty head.”

Either position is defensible. Say in your methodology chapter which one you took and how you guarded against reading your data through earlier theory. Memos and a reflexive journal are the natural evidence to point to.

One round of data isn’t enough

Because theoretical sampling needs analysis to guide collection, grounded theory needs more than one round of data collection.

If you’re collecting everything in one go, or analyzing a fixed dataset you didn’t collect, theoretical sampling becomes very hard. Glaser and Strauss allowed it in principle, by sampling within existing data, but only from “a large mass of data,” which few dissertation datasets are.

Building a grounded theory from published studies, a request we’ve seen supervisors make more than once, is possible in principle (Glaser and Strauss used published research to develop broader, “formal” theory). But it’s hard to do well in a dissertation: those findings have already been analyzed by someone else, and you can’t go back to their participants for more. In each case, a different method will usually serve you better.

Show the method, don’t just name it

“Data were analyzed using a grounded theory approach” tells your examiner nothing. In your methodology chapter, name your school, then show each practice: how you coded (with an excerpt, as both theses above do), how constant comparison changed your codes, how your sampling followed your analysis, what a memo looked like, and how you judged saturation. In your findings chapter, present the theory itself, its categories and how they connect, not a list of themes.

In our experience, the analysis procedure is often the first thing examiners probe in a grounded theory viva, because the method comes in so many versions. A chapter that shows each step answers most of those questions before they’re asked.

When grounded theory is the wrong choice

Grounded theory is the right design when you want to explain a process, there’s little existing theory about it, and you can collect data in rounds. It’s usually the wrong choice when:

• You’re applying or testing an existing framework. That’s deductive work. Use thematic analysis or qualitative content analysis with your framework as the lens.
• You only have a small fixed dataset, or one round of collection. You’ll have little room to sample theoretically, so consider a case study or thematic analysis.
• You want to describe experiences rather than explain a process. Phenomenology or thematic analysis will fit better.
• Your timeline is tight. Rounds of collection, analysis and ethics amendments take time a one-year master’s often doesn’t have.

None of these alternatives is a lesser method. Choosing the design that fits your question, and naming it accurately, will get you further than a prestigious label you can’t defend. If you’re still deciding, our guide to choosing a qualitative analysis method compares the main options side by side.

Still have questions?

Is grounded theory qualitative or quantitative?

Grounded theory is almost always used as a qualitative research design, with interviews, observations or documents as data. Its founders didn’t limit it that way: Glaser and Strauss’s 1967 book includes a chapter on generating theory from quantitative data. But in dissertations today, grounded theory means qualitative work in practice.

Who founded grounded theory?

Sociologists Barney Glaser and Anselm Strauss developed grounded theory and set it out in their 1967 book, The Discovery of Grounded Theory. They drew on their earlier study of how hospital staff handled dying patients. The two later disagreed about the method, and Kathy Charmaz then developed a constructivist version, which is why there are now three main versions.

What are the limitations of grounded theory?

Grounded theory takes time, because data collection and analysis happen in rounds. It comes in several competing versions, so you have to justify the one you choose. The classic and constructivist advice to delay the literature review clashes with what many programs require. And a dissertation’s theory usually explains one specific setting (what Glaser and Strauss called substantive theory), so it can’t be generalized the way a statistical finding can.

How many participants does grounded theory need?

There’s no fixed number. A grounded theory study keeps sampling until its categories are saturated, so the sample size is an outcome of the analysis rather than a target set in advance. For a sense of scale, a 2018 nursing PhD on diabetic foot ulcers interviewed 30 people, and a 2023 psychology PhD on multiracial identity worked with 23. Your committee will still want an estimate in your proposal, so give one and explain why it may change.

Can I use grounded theory with secondary data?

It’s possible, but it’s hard. Glaser and Strauss accepted theoretical sampling within previously collected data, but only from a large mass of it, because your analysis still has to be able to choose what to look at next. A small, fixed secondary dataset rarely allows that. If your study uses a limited set of existing documents or records, thematic analysis, content analysis or a case study design will usually fit better and be easier to defend.

How is constructivist grounded theory different?

Constructivist grounded theory, developed by Kathy Charmaz, treats a theory as something the researcher constructs through their interaction with participants and data, rather than something discovered in the data. It uses two main coding stages, initial and focused, and is more flexible than the classic and Straussian versions. It suits researchers who see themselves as part of the research rather than neutral observers.

What is a memo in grounded theory?

A memo is a note you write to yourself about your codes, categories and emerging ideas while you analyze. Charmaz calls memo-writing “the pivotal intermediate step between data collection and writing drafts of papers.” Memos are where you compare codes, question your interpretations, correct mistakes and build the relationships that become your theory.

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

Jansen, D. (2026). Grounded Theory: What It Is and How It Works. Grad Coach. https://gradcoach.com/grounded-theory/

Copy citationMore citation formats

• Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77-101.
• Charmaz, K. (2006). Constructing grounded theory: A practical guide through qualitative analysis. Sage.
• Costa, I. G. (2018). The journey toward engagement in self-care management of diabetic foot ulcer: A constructivist grounded theory study [Doctoral dissertation, Queen’s University].
• Dunne, C. (2011). The place of the literature review in grounded theory research. International Journal of Social Research Methodology, 14(2), 111-124.
• Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for qualitative research. Aldine.
• Kelle, U. (2005). “Emergence” vs. “forcing” of empirical data? A crucial problem of “grounded theory” reconsidered. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 6(2), Article 27.
• Kenny, M., & Fourie, R. (2015). Contrasting classic, Straussian, and constructivist grounded theory: Methodological and philosophical conflicts. The Qualitative Report, 20(8), 1270-1289.
• Sbaraini, A., Carter, S. M., Evans, R. W., & Blinkhorn, A. (2011). How to do a grounded theory study: A worked example of a study of dental practices. BMC Medical Research Methodology, 11, 128.
• Scarbrough, S. M. (2023). Understanding multiracial identity development using constructivist grounded theory [Doctoral dissertation, University of Hawaiʻi at Mānoa].
• Thornberg, R. (2012). Informed grounded theory. Scandinavian Journal of Educational Research, 56(3), 243-259.
• Vander Linden, K. L. (2026). Themes are not theory: Understanding theory development in grounded theory. Grounded Theory Review, 26, Article 19.