Grounded theory is a qualitative research methodology that builds new theory directly from data rather than testing a hypothesis set in advance. First articulated by sociologists Barney Glaser and Anselm Strauss in The Discovery of Grounded Theory (1967), it systematically generates an explanatory theory that is “grounded” in the evidence — data are collected and analysed at the same time, codes are compared constantly, and the emerging theory dictates what to sample next. Use it when little is known about a process and you want to explain how and why something happens, not merely describe it.
In short: you start with an open research question, gather data (usually interviews or observations), code it line by line, group codes into categories through constant comparison, and keep sampling until no new properties emerge — the point of theoretical saturation — arriving at a core category that ties the theory together.
What is grounded theory?
Grounded theory is both a methodology and a set of analytic procedures for developing theory inductively from qualitative data. Where most research moves deductively — from an existing theory, to a hypothesis, to data that confirm or refute it — grounded theory reverses the direction. It moves from data upward, abstracting patterns into concepts, concepts into categories, and categories into a coherent theory that explains the phenomenon under study. This inductive logic sits at the heart of the method; if you are unsure how it differs from the top-down alternative, our guide to inductive and deductive reasoning sets out the contrast clearly.
Glaser and Strauss developed grounded theory partly in reaction to the grand, abstract theorising that dominated mid-twentieth-century sociology. Their insistence was simple but radical: theory should earn its keep by fitting the data it claims to explain, and it should be discovered through disciplined analysis rather than imposed from the armchair. The defining commitment is that the researcher enters the field without a fixed hypothesis and lets the analytic categories emerge.
“Generating a theory from data means that most hypotheses and concepts not only come from the data, but are systematically worked out in relation to the data during the course of the research.” (Source: Glaser & Strauss, 1967)
Because the analysis is conducted alongside data collection rather than after it, grounded theory is an iterative, cyclical method. You do not collect everything first and analyse later; instead each round of analysis shapes the next round of data gathering. This is what makes the approach feel different in practice from, say, a one-pass thematic analysis, where coding typically follows a completed dataset.
The core procedures of grounded theory
Five interlocking procedures distinguish grounded theory from other qualitative approaches. None of them works in isolation — together they form the engine that turns raw data into theory.
1. Theoretical sampling
In grounded theory you do not fix your sample in advance. You begin with an initial group of participants, analyse what they tell you, and then deliberately choose the next participants or settings on the basis of what the emerging analysis still needs to clarify. If an early category — say, “managing uncertainty” — looks important but thin, you go and find people likely to illuminate it. Sampling is therefore driven by the developing theory, not by demographic representativeness. This is a fundamentally different logic from probability sampling, and it pairs naturally with the qualitative methods of data collection — in-depth interviews, focus groups and observation — that grounded theorists rely on.
2. The constant comparative method
This is the analytic heartbeat of grounded theory. Every new piece of data is compared with previous data and with the codes and categories already created: incident is compared with incident, incident with category, and category with category. Constant comparison forces the analyst to ask repeatedly, “How is this similar to or different from what I have already seen?” The discipline of comparison is what sharpens fuzzy first impressions into precise, well-defined categories with clear properties and dimensions.
3. The three coding stages
Grounded theory analysis classically proceeds through three coding stages, each more abstract than the last:
- Open coding — breaking the data into discrete parts and labelling them with conceptual codes, line by line or incident by incident. The aim is to stay close to the data and generate as many codes as the material warrants.
- Axial coding — reassembling the fractured data by identifying relationships between codes, grouping them into categories and sub-categories, and specifying conditions, contexts and consequences.
- Selective coding — integrating and refining the categories around a single core category that accounts for most of the variation in the data and ties the theory together.
The figure below shows how these stages connect to the wider cycle of data collection and theory building, including the feedback loop created by constant comparison and theoretical sampling.
4. Memo-writing
Memos are the researcher’s running analytic notes — informal, dated records that capture ideas about codes, hunches about relationships, and questions to pursue. Glaser regarded memo-writing as so central that he urged analysts to stop coding and write a memo the moment an idea strikes. Memos are where the theory is actually thought through; the codes are only raw material. Skipping memos is one of the most damaging shortcuts a novice can take, because the theoretical insight lives in the memos, not in the code list.
5. Theoretical saturation
Saturation is the stopping rule. You keep collecting and analysing data until additional cases yield no new properties, dimensions or relationships for your categories — the categories are “saturated”. Saturation, rather than a pre-set sample size, determines when data collection ends. In practice this is judged category by category, and a well-conducted study reports how saturation was reached.
The main variants of grounded theory
Grounded theory is not a single fixed recipe. After Glaser and Strauss’s original collaboration, the method split into distinct schools that disagree about the researcher’s stance, the use of prior literature, and how prescriptive the coding should be. The three most influential versions are summarised below.
| Feature | Glaserian (classic) | Straussian | Constructivist (Charmaz) |
|---|---|---|---|
| Key figures | Barney Glaser | Strauss & Corbin | Kathy Charmaz |
| Philosophical stance | Objectivist / positivist-leaning; theory “emerges” | Post-positivist; more structured | Constructivist / interpretivist; data are co-constructed |
| Role of the researcher | Neutral discoverer; minimise preconceptions | Active analyst applying a coding framework | Co-producer of meaning; reflexivity central |
| Coding approach | Open, then selective; theoretical codes | Open, axial (coding paradigm), selective | Initial, focused, then theoretical coding |
| Prior literature | Delay review to avoid “forcing” | Literature can inform sensitising concepts | Used reflexively; no claim of a blank slate |
| View of the theory | Discovered in the data | Systematically built from the data | Constructed between researcher and participants |
For a dissertation, the practical message is to choose one version and apply it consistently. Mixing Glaser’s “emergence” rhetoric with Strauss and Corbin’s axial-coding paradigm, then claiming Charmaz’s reflexivity, will read as muddled. State your chosen variant in the methodology and justify it against your epistemology.
Grounded theory step by step
Although the variants differ, a generic grounded theory study follows a recognisable sequence. Treat the steps as a cycle, not a one-way pipeline — you will loop back repeatedly.
- Frame an open research question. Ask “how” or “what is going on here?” about a process, rather than stating a hypothesis to test.
- Collect an initial slice of data. Usually a small number of interviews or observations to get started.
- Open-code immediately. Label the data line by line; do not wait to gather everything first.
- Write memos. Record every analytic idea as it arises, and date it.
- Compare constantly. Set each new incident against earlier ones; merge, split and rename codes as patterns sharpen.
- Sample theoretically. Let the emerging categories tell you whom or what to study next.
- Move to axial coding. Link codes into categories, specifying their properties, conditions and consequences.
- Identify the core category. Through selective coding, find the central concept that integrates the others.
- Continue until saturation. Stop collecting data when new cases add nothing to the categories.
- Write up the theory. Present the integrated theory, illustrated with data extracts and your memos’ reasoning.
A worked example: coding interview data
The best way to see grounded theory in action is to watch raw data climb the ladder of abstraction — from codes, to categories, to a core category. The example below uses a short interview excerpt from a hypothetical education study on first-generation university students. If your own data come from interviews, the practical work of preparing them well begins with careful transcription; our guide on how to transcribe an interview is a useful companion before any coding starts.
“Nobody in my family went to uni, so I didn’t know you could just email a lecturer. I worked it out by watching what the others did and copying them. I felt like I was always a step behind, but I didn’t want to ask and look stupid.”
Step 1 — Open coding (line by line):
- “Nobody in my family went to uni” → code: no family precedent
- “didn’t know you could email a lecturer” → code: missing institutional knowledge
- “watching what the others did and copying them” → code: learning by observation
- “always a step behind” → code: perceived lag
- “didn’t want to ask and look stupid” → code: fear of exposure
Step 2 — Axial coding (group codes into categories):
- no family precedent + missing institutional knowledge → category: knowledge gap
- learning by observation → category: covert coping strategies
- perceived lag + fear of exposure → category: managing a stigmatised identity
Step 3 — Selective coding (find the core category): After comparing Aisha’s account with twenty further interviews and sampling more students theoretically, the same pattern recurs — students decode unfamiliar university norms privately to avoid being seen as not belonging. The integrating core category becomes “decoding belonging in secret”, which explains the relationship between the knowledge gap, the covert strategies and the identity work. That core category — not any single interview — is the theory the study contributes.
Notice that the data drove every upward move; no category was decided in advance. This bottom-up trajectory is what people mean when they call grounded theory a qualitative data analysis method that generates rather than applies theory.
When should you use grounded theory?
Grounded theory is a strong choice in specific circumstances and a poor one in others. Reach for it when the following apply:
- The topic is under-researched and there is no adequate existing theory to test.
- Your interest is in a social or psychological process — how something unfolds, changes or is managed over time.
- You can collect and analyse data iteratively, returning to the field as your categories develop.
- You want an explanatory theory as your end product, not just a description or a set of themes.
- You are prepared to keep an open mind and let the data, rather than your assumptions, lead.
Conversely, grounded theory is the wrong tool if you already have a theory you wish to confirm, if you need statistical generalisation to a population, or if your timetable cannot accommodate iterative, open-ended fieldwork.
Strengths and limitations
Like every methodology, grounded theory trades certain advantages against real costs. Weighing them honestly will strengthen your methodology chapter.
| Strengths | Limitations |
|---|---|
| Generates rich, original theory that fits the data closely | Time-consuming and labour-intensive; iterative cycles take months |
| Well suited to new or poorly understood phenomena | Procedurally complex; easy to apply superficially or inconsistently |
| Systematic and transparent — the audit trail of codes and memos is explicit | Theoretical sampling and saturation are hard to judge and to defend |
| Keeps the researcher close to participants’ own meanings | Researcher influence is unavoidable; complete neutrality is a myth |
| Flexible across disciplines — psychology, business, education, health, sociology | Findings are not statistically generalisable to wider populations |
The two recurring criticisms worth pre-empting are researcher influence and analytic complexity. The classic Glaserian claim that theory simply “emerges” from data, untouched by the analyst, is now widely rejected; Charmaz’s constructivist answer — acknowledge your role and document your reflexivity — is the more defensible position for most modern dissertations.
Struggling to apply grounded theory in your dissertation?
Our subject-matter experts can help you design the study, code your data and build a defensible theory chapter.
Common mistakes to avoid
Most weak grounded theory studies fail in predictable ways. Steer clear of these traps:
- Forcing the data. Imposing pre-existing concepts or a favoured theory onto the data, instead of letting categories emerge through comparison. This is the cardinal sin Glaser warned against — it produces a theory that fits your expectations, not the evidence.
- Skipping memo-writing. Treating coding as the whole job and never writing analytic memos. The theoretical reasoning then has nowhere to live, and the final “theory” is just a tidied code list.
- Fixing the sample in advance. Recruiting a set number of participants up front defeats theoretical sampling and usually leaves categories under-developed.
- Claiming saturation without evidence. Asserting that categories are saturated because you ran out of time, rather than because new data genuinely added nothing.
- Describing instead of theorising. Stopping at rich themes and never integrating them around a core category — which leaves you with thematic analysis, not grounded theory.
- Mixing the variants incoherently. Borrowing procedures from Glaser, Strauss and Charmaz without acknowledging that they rest on different philosophical assumptions.
Doing grounded theory well
A credible grounded theory study is recognisable by a few hallmarks: an open question, simultaneous data collection and analysis, a visible trail of codes and memos, theoretical (not convenience) sampling, an explicit account of how saturation was judged, and a clearly stated core category that integrates the theory. Declare your chosen variant early, stay reflexive about your own influence on the data, and let the evidence — not your assumptions — have the final word. Do that, and grounded theory will reward you with something genuinely rare in student research: a new, defensible explanation that is truly grounded in what your participants told you.
