Grounded theory: what it is and when to use it

Grounded theory is among the most cited qualitative methodologies in graduate research and probably the most misapplied. The name gets attached to any study that codes interview transcripts, which is roughly like calling any study with numbers an experiment. Actual grounded theory is a specific, demanding package: a way of generating new theory from data through cycles of coding, comparison, memo writing, and targeted sampling.

Done properly, it produces something rare in a thesis: an original theoretical contribution. Done loosely, it produces a thematic analysis with the wrong label, and examiners who know the methodology will say so. This guide explains what grounded theory actually is, how the three schools differ, when it is the right choice, and how to execute the mechanics: coding, theoretical sampling, saturation, and memos.

What grounded theory actually is

Grounded theory inverts the standard research sequence. Instead of deriving hypotheses from existing theory and testing them against data, you start with data and build theory out of it, staying close to what participants say and do until an explanatory framework earns its way into existence. The output is not a list of themes. It is a theory: a set of related concepts that explains a process, with a core category at its centre and stated relationships among the parts. If your final chapter cannot be drawn as a diagram with labelled links, you have categories, not yet a theory.

Three commitments define the method regardless of school. Constant comparison: every new piece of data is compared against existing codes, and every code against the others, continuously rather than in a single pass at the end. Theoretical sampling: after initial analysis, you decide what data to collect next based on what the developing theory needs. And memo writing throughout, which is where the actual theorising happens. If a study did none of these, it was not grounded theory, whatever the methods section says.

The three schools of grounded theory

Grounded theory split into schools early, and your thesis must pick one and stay consistent, because the schools disagree about coding procedure and about what the researcher brings to the data.

• Classic (Glaserian) grounded theory: the most inductive version. The researcher approaches data with as few preconceptions as possible and trusts emergence, coding openly and then selectively around a discovered core category.
• Straussian grounded theory (Strauss and Corbin): the most procedural version. It adds structured tools, notably axial coding with a coding paradigm that relates categories through conditions, actions, and consequences. Popular in theses because the steps are teachable and auditable.
• Constructivist grounded theory (Charmaz): the most reflexive version. It treats theory as co-constructed by researcher and participants rather than discovered, uses initial and focused coding with gerunds to keep the analysis on processes, and expects an explicit account of the researcher's position.

Pick a school and stay consistent

Committees rarely mind which school you choose. They mind mixing them unknowingly: citing one school's coding stages while performing another school's procedure is one of the first inconsistencies a methodologically literate examiner checks for. State your school in the methods chapter, cite its canonical sources, and let its vocabulary run through the whole thesis, because a Straussian axial coding section sitting inside a Charmaz-cited design reads as confusion, not synthesis. If you must borrow a tool across schools, borrow knowingly: name the borrowing, cite a precedent for it, and explain why your study needs it.

When to use grounded theory (and when not to)

Grounded theory fits a specific situation:

• The phenomenon is a social process: how people move through, manage, or make sense of something over time.
• Existing theory is absent, thin, or imported from a context where it does not obviously apply.
• Your question asks how or why a process unfolds, not how often or how much.
• You can return to the field repeatedly, because theoretical sampling requires collecting data in more than one round.

The honest alternative when it does not fit

Grounded theory does not fit when a strong theory already maps your territory (then you are testing or extending, not generating), when your question is descriptive rather than process-oriented, or when data collection is locked to a single round by access or ethics constraints, which quietly removes theoretical sampling from the design.

Be honest about that last case. A one-round interview study analysed with grounded theory techniques can be excellent research, but consider calling the approach a grounded theory informed thematic analysis rather than claiming the full methodology. Examiners punish mislabelled methodology far more severely than modest methodology, and the relabelling costs you nothing except a smaller claim.

Grounded theory coding: open, axial, and selective

In the Straussian sequence, open coding fractures the data: line by line or incident by incident, you label what is happening, generating many provisional codes and gradually grouping them into categories. Axial coding reassembles: you take a category and relate its subcategories through the coding paradigm, asking under what conditions it appears, what actions and interactions constitute it, and what consequences follow. Selective coding integrates: you commit to a core category, the one that explains the most variation in the data, and systematically relate everything else to it.

The constructivist sequence is leaner: initial coding (fast, provisional, close to the data, coded with gerunds like managing disclosure to keep processes in view), then focused coding, where the most frequent and significant initial codes are tested against the full dataset and elevated into the analytic frame. Whichever sequence you use, the discipline is the same: codes must earn their place through constant comparison, and the write-up must show the lineage from raw data to category to theory. Keep a codebook with definitions and example excerpts as you go; it becomes an appendix examiners genuinely read.

Theoretical sampling, saturation, and memo writing

Theoretical sampling is the feature that most distinguishes grounded theory from other qualitative designs, and the one most often skipped. After your first round of analysis, the developing categories will have gaps: conditions you have not observed, variations you cannot explain, comparisons you need. You sample to fill those gaps, choosing the next participants, sites, or documents because of what the theory needs, not because of a sampling frame fixed in advance. Your methods chapter should narrate at least one concrete instance: what the category lacked, whom you therefore sought, and what changed.

Saturation, in its grounded theory sense, is theoretical: categories are saturated when new data no longer adds properties or forces revisions, not merely when interviews start sounding similar. Claiming saturation obliges you to show the machinery that detected it, which is what memos are for. A bare sentence asserting that saturation was reached convinces no one. A memo trail showing category development flattening does.

Memos are the working notes in which you argue with your own codes: why this category, how it relates to that one, what it fails to explain, what to sample next. They are written continuously from the first interview to the final draft, and they are the evidence that theorising happened. In a viva, the memo trail is your audit trail, showing the theory being built rather than asserted.

Practical habits that pay off: date every memo, title it with the category it develops, write in full sentences because fragments do not survive six months, and sort memos by category when you begin writing up. A good memo bank means the analysis chapters are partly drafted before you start them.

The literature review problem in grounded theory

Classic grounded theory advised delaying the literature review so that existing concepts would not contaminate emergence. A thesis cannot fully obey: your proposal needs a review, and your examiners expect command of the field. The workable position, and the one constructivist grounded theory endorses, is to review the literature early but hold it consciously at arm's length during coding, then return to it late in the analysis and treat it as another source to compare your theory against.

Document that stance explicitly in the methods chapter so the sequencing reads as design rather than negligence. Keeping the field's papers organised in the Library with reading statuses makes the two-phase engagement manageable, and the mechanics of an AI-assisted review that stays grounded in real sources are covered in How to write a literature review with AI.

Writing up a grounded theory thesis

The findings chapters of a grounded theory thesis present the theory, not a tour of themes. Lead with the core category and the storyline: one or two paragraphs stating the whole theory plainly. Then develop each major category in turn, with its properties, its relationships to the core, and the data that earned it, quoting enough raw material that the reader can check the grounding. Diagrams of the category structure help examiners more than prose alone. Readers should be able to open any findings chapter and see the same theory from a different side, rather than a fresh set of themes.

Late in the write-up you position the theory against the literature you held at arm's length: where it converges with existing accounts, where it extends them, and what it explains that they do not. An evidence matrix in the Synthesis Lab lines your categories up against published findings paper by paper, which turns the positioning section from hand-waving into an auditable comparison. Grounded drafting in the Thesis Editor helps for the same reason: citations resolve to real papers and claims are checked against full text, so the theoretical positioning cannot quietly drift beyond what the sources say.

Common grounded theory pitfalls

The recurring failures examiners see:

• Thematic analysis wearing a grounded theory label: no constant comparison, no theoretical sampling, no theory at the end.
• School mixing: Straussian axial coding cited alongside Glaserian emergence talk, with no acknowledgement of the tension.
• A core category that is a topic, not a process: communication explains nothing, negotiating disclosure can.
• Saturation asserted rather than evidenced.
• No memos, or memos written retrospectively for the appendix.
• A findings chapter with rich description and no stated relationships among categories, which means no theory.

Is grounded theory worth it for your thesis?

Grounded theory is a heavy methodology, and the weight is the point: the procedures exist to make an inductive claim defensible. If your project genuinely needs new theory about a process, commit to a school, run the machinery honestly, and the result can be the strongest contribution a qualitative thesis can make. If what your project actually needs is a well-executed thematic analysis, claim that instead, execute it properly, and spend the saved effort on the analysis itself. Either path can pass with distinction. The path that fails is the one where the label and the practice do not match.