What grounded theory actually commits you to, the three versions you must choose between, and why most theses claiming it are doing something else.
Grounded theory is a methodology for building an explanatory theory upwards from data rather than testing one downwards. It is not a way of analysing interviews. It commits you to collecting and analysing at the same time, to letting the developing theory decide who you approach next, and to continuing until that theory stops changing. If you collected a fixed set of interviews and then coded them, you have done thematic analysis — which is a perfectly respectable thing to have done, and should be called by its own name.
Grounded theory is defined by a small number of practices that have to be present together. Take any one away and what remains is qualitative analysis of some other kind.
Simultaneous collection and analysis. You analyse the first interviews before conducting the next. This is not a scheduling preference — it is what makes everything else possible, because the analysis has to be far enough along to direct what you do next.
Theoretical sampling. Who you approach next is decided by what the developing theory needs, not by a recruitment plan fixed in advance. If a category is thin, you go looking for people or situations likely to fill it out. This is the practice most often absent in theses claiming the method, and it is the one an examiner will probe first.
Constant comparison. Every new incident is compared with those already coded, and every code with every other. Categories are built and refined by comparison rather than assembled at the end.
Memo writing. Analytic notes written throughout are where the theory is actually developed. Memos are not a record of the analysis; they are the analysis, and they become the raw material of the eventual write-up.
Theoretical saturation. You stop when further sampling stops changing the theory — which only means anything if you were sampling theoretically in the first place.
A core category. The endpoint is an integrated explanation organised around a central concept, not a set of themes. If your output is a list of themes, you have not produced a grounded theory.
Grounded theory split into distinct schools that disagree on questions your examiner may well care about. Naming which you followed, and citing it, is expected.
Following Glaser, who with Strauss originated the method. Emphasises that the theory should emerge from data with as little preconception as possible, resists prescriptive coding procedures, and is firm that the researcher should not go into the field with a framework in mind. Includes the well-known and, for doctoral candidates, awkward position that a substantive literature review should be delayed until the theory has taken shape.
Following Strauss and Corbin, whose divergence from Glaser produced a public split in the field. More structured and more teachable: explicit coding stages — open, axial and selective — and an analytic frame for relating categories to conditions, actions and consequences. The most commonly used version in doctoral work, partly because the procedures can be described in a methods chapter.
Following Charmaz. Accepts the core practices while rejecting the idea that theory is discovered by a neutral observer: theory is constructed through the interaction between researcher and participants, and the researcher’s position is part of the account rather than something to be eliminated. Requires reflexivity as an explicit component, and sits more comfortably with contemporary qualitative thinking than the discovery language of the classic version.
Terminology varies between versions, which is itself a reason to name yours. In the Strauss and Corbin vocabulary most commonly taught:
Charmaz uses a different scheme — initial and focused coding, with theoretical coding to relate categories — and is explicit that axial coding as a formal procedure is optional. Borrowing terms across versions is a small error that signals the literature has not been read closely.
Whichever you follow, these are not sequential stages completed once. You will be open coding new data while selectively coding older material, because collection and analysis are running together.
This trips up almost every doctoral candidate using grounded theory, and it is better to face it early.
Classic grounded theory holds that engaging deeply with the literature before analysis contaminates the theory with existing concepts. But a doctoral programme requires a proposal with a literature review, an ethics application specifying what you will do, and often an upgrade examination — all before you collect anything. The methodological ideal and the institutional requirement are in direct conflict.
Workable ways through, all of which have been used successfully:
What does not work is claiming classic grounded theory while presenting a conventional front-loaded literature review, and hoping nobody notices the contradiction.
It suits questions about social or psychological processes — how something develops, how people manage a situation over time, what accounts for variation in how a problem is handled — particularly where existing theory is thin.
It suits badly: questions about the meaning of a single experience, where phenomenology fits better; descriptive questions about what people think, where thematic analysis is more honest; anything with a fixed sample agreed in advance with a gatekeeper, since theoretical sampling then becomes impossible; and projects with hard deadlines, because you cannot know at the outset how many participants the theory will require.
That last constraint is practical and severe. Ethics committees generally want a defined number, funders want a plan, and grounded theory cannot honestly supply either. Negotiating an approved range, with a stated rationale, is the usual accommodation — and worth doing before the proposal is submitted rather than after.
A methodology for constructing an explanatory theory from data rather than testing an existing one. It requires collecting and analysing simultaneously, sampling further participants on the basis of what the developing theory needs, constant comparison between incidents and categories, memo writing throughout, and continuing until the theory stops changing. The output is an integrated theory organised around a core category, not a set of themes.
Choosing who or what to study next on the basis of what your developing analysis requires, rather than following a recruitment plan fixed in advance. If a category is underdeveloped, you deliberately seek out people or situations likely to develop it. It is the practice most often missing from theses that claim grounded theory, and its absence is usually the first thing an examiner asks about.
Grounded theory is a whole methodology governing sampling, collection and analysis together, and aims at an explanatory theory. Thematic analysis is an analysis method applied to data however it was collected, and aims at patterns of meaning. If you recruited a fixed sample, collected all the data and then coded it, that is thematic analysis. Calling it grounded theory because you used open coding is the most common mislabel in doctoral qualitative work.
Most doctoral candidates find Strauss and Corbin’s systematic version easiest to describe and defend, because its procedures can be written down clearly. Charmaz’s constructivist version fits better if you are uncomfortable with the idea of theory being discovered by a neutral observer, and it handles prior knowledge more realistically. Glaser’s classic version is the most demanding to reconcile with doctoral requirements. Whichever you choose, name it and cite it — the versions disagree, and examiners know it.
This is a genuine conflict, since classic grounded theory advises delaying it while your university requires one in the proposal. Common solutions are to follow Charmaz, who treats prior knowledge as unavoidable and manages it reflexively; to conduct an initial review establishing the need for the study and defer detailed engagement until the theory has formed; or to treat the literature as further data compared against your categories later. Any of these is defensible if you explain it. Claiming the classic version alongside a conventional front-loaded review is not.
It cannot be known in advance, because sampling is driven by the developing theory and stops when that theory stops changing. That creates a real problem with ethics committees and funders who need a number, and the usual accommodation is to propose a justified range and explain the sampling logic. Any figure quoted as typical is a convention rather than a requirement.
Grounded theory is difficult to claim convincingly after the fact, because theoretical sampling has to happen while you collect. Describe your question and your access, and a PhD in your field will tell you whether it is workable.
Leave an email or a WhatsApp number — whichever you prefer — and tick how we should reach you. We reach out within 30 minutes.
