Thematic Analysis vs. Grounded Theory: Which Qualitative Method Belongs in Your Dissertation

The single question I hear most often from doctoral candidates who are finalising Chapter 3 is some version of this: should I be doing thematic analysis or grounded theory? The question almost always arrives late — after the IRB application has been filed, sometimes after data collection has begun — and it arrives with a particular kind of anxiety that is different from other methodology questions. The candidate has read both. They have heard colleagues use both terms. They have seen committee members use them interchangeably in feedback that now sits in their revision notes. They are not confused about what the methods are. They are confused about what it would mean to be wrong about which one they chose, because at the stage they are asking the question, being wrong has real consequences for the shape of the chapter, the timeline to completion, and what the committee is actually going to say at the defense.

The question matters because the answer is not preferential. This is the thing most dissertation-methods courses do not say clearly: thematic analysis and grounded theory are not two flavours of the same approach. They rest on different epistemological commitments. They require different coding procedures. They produce different deliverables. A study designed for thematic analysis cannot simply be reframed as grounded theory by changing the vocabulary in Chapter 3, and a genuine grounded theory study cannot be compressed into a thematic analysis without abandoning the core methodological commitment that makes the approach coherent. The committee member who reads a Chapter 3 that claims to be doing grounded theory but is structured like a thematic analysis will catch it at the defense — and catching it at the defense is a revision that takes months, not weeks.

This article walks through five things. The first is the epistemological divide: how each method understands knowledge and what that commitment requires of the researcher before a single code is written. The second is the coding flow, side by side — the Braun and Clarke six-phase sequence for thematic analysis against the open-axial-selective sequence for grounded theory, with attention to where the two diverge in ways that change the timeline of your study. The third and fourth are the decision criteria for each method: when to pick thematic analysis, when to pick grounded theory, and — critically — when not to pick each one. The fifth is the common mistakes: two per method, the ones that convert a defensible Chapter 3 into a revision spiral. The goal is that by the end of this article you know which method belongs in your study, and you know it for reasons your committee cannot argue with.

The Epistemological Divide: How Each Method Understands Knowledge

Thematic analysis, as codified by Braun and Clarke in their foundational 2006 paper and elaborated in the decade since, is a method without a fixed epistemological home. This is its most important characteristic and the one most candidates miss. Thematic analysis can be done from a realist position — the themes the researcher identifies reflect something real and stable in the participants' experience of the world — or from a constructivist position — the themes are a co-construction between the participant's account and the researcher's interpretive frame. The method does not dictate which. What the method requires is that the researcher declares which, clearly and explicitly, in Chapter 3, and that the declared position is consistent with the research question, the data collection approach, and the analytic choices made throughout the study. The candidate who does not declare a positionality is not doing thematic analysis — they are doing something that looks like thematic analysis from the outside but that has no methodological spine. That study will not survive a committee defense.

Grounded theory does not offer the researcher that flexibility. Grounded theory — whether in the Strauss and Corbin procedural tradition or in Kathy Charmaz's constructivist reformulation — begins from a specific and non-negotiable epistemological premise: theory does not pre-exist the data. A grounded theory study is not a study that applies an existing theoretical framework to new data. A grounded theory study is a study in which the theory emerges from the data through an iterative process of collection, coding, and constant comparison, and the study is not complete until that theory has been built and tested through theoretical sampling to a point of saturation. A researcher who enters a grounded theory study with a theoretical framework already declared in Chapter 2 has, in methodological terms, already violated the core commitment of the approach. The committee member trained in grounded theory will identify this immediately, and the defense conversation will be about the fundamental incoherence of the design rather than the quality of the data.

The practical implication of this divide is visible at the committee table before the defense begins. In a thematic analysis, the researcher's role, the researcher's positionality, the researcher's declared relationship to the data — these are methodological commitments that belong in Chapter 3 and that will be questioned there. In a grounded theory study, the researcher's role matters differently: the grounded theorist must be able to demonstrate that they did not bring a pre-formed theory to the data, that the theoretical sampling was genuinely driven by what the data was not yet explaining, and that saturation was a methodological decision rather than an IRB-sample-completion event. These are different kinds of accountability. Confusing which one your study requires means writing Chapter 3 against the wrong accountability standard, and writing Chapter 3 against the wrong accountability standard means rewriting Chapter 3 before you can defend.

The Coding Flow: Step by Step, Side by Side

The coding sequences for the two methods look similar from a distance. Both involve reading data carefully, assigning codes to segments of text, and moving from initial codes toward more abstract analytic categories. The similarity ends there. The structure, the sequencing, and — critically — the data collection loop are fundamentally different, and the difference has direct consequences for how long the study takes and what kind of dissertation it produces.

Thematic Analysis: The Braun and Clarke Six Phases

Braun and Clarke describe thematic analysis as a six-phase process, and the sequence matters. Phase one is familiarisation: the researcher reads the entire dataset — all transcripts, all field notes, all documents — multiple times, noting initial ideas but not yet coding. The goal is immersion in the data before interpretation begins. Phase two is generating initial codes: the researcher works systematically through the dataset, generating initial codes for features of the data that appear interesting or relevant to the research question. At this stage, codes are applied liberally; the researcher is not yet discriminating between what will become a theme and what will not. Phase three is searching for themes: the researcher collects all codes and considers how they might combine into broader themes. Codes are sorted, clustered, and grouped into candidate themes. Phase four is reviewing themes: the researcher tests each candidate theme against the data — does this theme tell a coherent story? Does the data actually support it? Some themes collapse, some split, some are abandoned. Phase five is defining and naming themes: each surviving theme is given a name and a precise definition that captures what the theme is about and what it is not about. Phase six is writing the report: the analysis is written as a narrative that weaves data extracts into the argument.

The data collection in a thematic analysis study is complete before phase one begins. The researcher is not returning to the field to collect additional data because the analysis demands it. The IRB-approved sample is the sample, and the analysis works within it. This is the structural feature that makes thematic analysis compatible with the standard dissertation timeline and with most IRB approval frameworks.

Grounded Theory: Open Coding, Axial Coding, Selective Coding, and the Saturation Loop

Grounded theory's coding sequence begins with open coding: the researcher reads the first wave of data and codes it at a granular, line-by-line level, asking what is happening here? without importing categories from prior theory. Open coding produces a large number of codes, many of them overlapping, some of them contradictory. The researcher is not attempting resolution at this stage — the goal is density of coverage. Axial coding follows: the researcher identifies relationships between open codes, building categories and sub-categories and identifying how they connect — conditions, contexts, actions, consequences. Selective coding is the third phase: one core category is identified that integrates all the others into a coherent account of the central social process the study is investigating. The selective code becomes the basis for the emerging theory.

What distinguishes grounded theory from thematic analysis at a structural level is the loop that follows selective coding: theoretical sampling. When the emerging theory has gaps — categories that are underdeveloped, connections that are asserted but not supported — the researcher returns to the field to collect additional data specifically targeted at filling those gaps. This is not a flaw in the original data collection plan. It is the method working as designed. Theoretical sampling continues until saturation: the point at which additional data no longer produces new theoretical insight. Saturation is a methodological judgment, not a sample-size criterion. A grounded theory study that stops at the IRB-approved sample size because the IRB approved that sample size — rather than because saturation has been reached — is a study that has not completed its methodological commitment. This distinction matters enormously at the defense.

When to Pick Thematic Analysis (and When You Shouldn't)

Thematic analysis is the right choice when the research question is descriptive or interpretive in nature — when the goal is to understand what a group of people think, feel, or experience about a phenomenon, and to describe the patterns in that experience systematically. It is the right choice when the relevant literature is already developed enough that the study does not need to build theory from scratch: there is a conceptual vocabulary in the field, the research question is situated within an existing conversation, and the contribution is to extend or complicate that conversation with new empirical data. Thematic analysis is also the right choice when the committee's methodological expectations are constructivist or interpretivist but not theoretical-generative — when the committee wants to know what the participants said and how the researcher interpreted it, not what new theory the study produced.

The four concrete decision criteria that point toward thematic analysis are: (1) the research question asks about experience, perception, or meaning rather than about a social process to be theorised; (2) the relevant literature is developed enough that a theoretical framework is already declared in Chapter 2; (3) the IRB-approved sample is fixed and there is no institutional pathway for additional data collection rounds if the analysis demands them; and (4) the dissertation timeline does not accommodate the open-ended iterative collection loop that grounded theory requires. If all four conditions hold, thematic analysis is the method. If any one of them is in tension with the study design, the tension is worth examining before the IRB application is filed.

Thematic analysis is not the right choice when the research question is explicitly asking for a new theory. The candidate who writes a research question beginning with what theory can explain... or how does this process work in a way that existing theory does not account for... has written a research question that calls for grounded theory, and choosing thematic analysis to answer it produces a Chapter 3 that cannot support the Chapter 5 discussion the question is pointing toward. Thematic analysis is also not the right choice when the committee has expertise in grounded theory and the phenomenon under study is genuinely under-theorised in the literature — a committee that expects a theoretical contribution will not accept a thematic description as a substitute, regardless of how well the analysis is done.

When to Pick Grounded Theory (and When You Shouldn't)

Grounded theory is the right choice when the research question is genuinely theoretical — when the goal is not to describe a phenomenon but to explain it, to build a theory that accounts for how a social process works in a way that existing theory does not. It is the right choice when the literature review in Chapter 2 reveals that the field does not yet have a coherent theoretical account of the phenomenon, when the gap in the literature is a theoretical gap rather than an empirical one. It is also the right choice when the committee has grounded theory expertise and the study is positioned as a theory-generating contribution — when the methodological expectations the candidate is working against are explicitly grounded in the Strauss-Corbin or Charmaz tradition.

The four concrete decision criteria that point toward grounded theory are: (1) the research question asks how or why a social process works, not what participants experience; (2) the literature is underdeveloped enough that a theoretical framework cannot be declared in Chapter 2 without importing a framework from an adjacent field that does not quite fit; (3) the IRB approval framework and institutional timeline can accommodate additional data collection rounds driven by theoretical sampling; and (4) the committee's methodological orientation is theory-generative and the dissertation is expected to produce a theoretical contribution.

Grounded theory is not the right choice when the dissertation timeline is fixed and the candidate cannot absorb the cost of additional data collection rounds. It is not the right choice when the committee is constituted primarily of post-positivists who will read the theory-building ambition of a grounded theory study as overreach for a dissertation-level project. And it is decisively not the right choice when the candidate has already declared a theoretical framework in Chapter 2 — not because the declared framework disqualifies the candidate from using GT vocabulary, but because a study with a pre-declared theoretical framework has already violated the core epistemological commitment that makes grounded theory coherent. A committee that knows grounded theory will call this out. The candidate who switches to thematic analysis at that point, reconciles the framework declaration with a constructivist positionality, and redesigns the study as a theoretically situated interpretive investigation will defend a more coherent Chapter 3 than the candidate who insists on the GT label against an incompatible design.

Common Mistakes

Four mistakes account for the majority of the Chapter 3 revision spirals I see in candidates who have already collected data. Two are thematic analysis mistakes. Two are grounded theory mistakes. None of them are errors about the methods themselves — the candidate almost always understands what thematic analysis is and what grounded theory is. They are errors about the gap between declared method and actual practice, and that gap is what the committee finds at the defense.

TA Mistake 1: Labelling codes as themes before looking for cross-data patterns. This is the most common thematic analysis error and the hardest one to catch before the defense because it produces results that look like themes. The error is in the phase sequence. A code is a label assigned to a segment of data that captures something about its content. A theme is a pattern that appears across multiple data sources — across participants, across transcripts, across contexts — and that represents something analytically significant about the phenomenon. A candidate who names codes in phase two and then promotes them directly to themes in phase five, without genuinely completing the searching-and-reviewing phases three and four, produces a set of themes that are, methodologically, just well-labelled codes. The committee question that surfaces this is: how did you determine that this was a theme rather than a code? If the answer is that the candidate labelled it a theme because it appeared frequently, the committee has found the error. Frequency is not thematic significance. Cross-data pattern is thematic significance.

TA Mistake 2: Failing to declare a positionality that matches the analytic approach. As discussed in the epistemology section, thematic analysis requires a declared positionality. The error is not in failing to declare one — most candidates know they need a positionality statement in Chapter 3. The error is in declaring a positionality that is inconsistent with the analytic choices made in the study. The candidate who declares a realist positionality and then produces themes that are explicitly interpretive — that represent the researcher's construal of what the data means rather than an account of what participants reported — has a Chapter 3 in which the declared positionality and the actual analysis are telling two different methodological stories. The committee that reads carefully will find the inconsistency. The fix is not to change the positionality statement. The fix is to understand what the declared positionality requires of the analysis and to write the analysis accordingly.

GT Mistake 1: Stopping data collection at the IRB sample ceiling rather than at saturation. This is the most consequential grounded theory error because it is structural, not cosmetic. A grounded theory study that stops collecting data because the IRB-approved sample is complete — rather than because theoretical saturation has been reached — is a study that has not completed the method. The theory that emerges from an incomplete grounded theory study will have gaps that the committee's theoretical sampling questions will find. The defense question is: how do you know you have reached saturation? The candidate who answers by citing the IRB sample size has told the committee that they stopped collecting data for an administrative reason, not a methodological one. The right answer describes the theoretical sampling process — what gaps were identified, what additional data was collected to address them, and how the candidate determined that additional data was producing no new theoretical insight. If that process did not happen, the study is not a grounded theory study in the methodological sense, and the candidate is better served by reframing it as a qualitative descriptive study than by defending a GT label the analysis cannot support.

GT Mistake 2: Using GT vocabulary in a study that is doing thematic analysis. Open coding, axial coding, selective coding, theoretical sampling, constant comparison, saturation — these terms belong to grounded theory. They are not interchangeable with the Braun and Clarke six phases, and they are not generic qualitative vocabulary that can be applied to any inductive analysis. The candidate who has done thematic analysis but has described it in GT terms — because GT terms sound more rigorous, because the committee mentioned GT in a feedback meeting, because a prior draft of Chapter 3 was written around a GT design that was later abandoned — has written a Chapter 3 that uses a vocabulary inconsistent with the method. A committee member who knows grounded theory will test the candidate's use of these terms against the actual study design, and the inconsistency will be visible. The fix is straightforward: use the method's own vocabulary. Braun and Clarke's terminology is precise, well-established, and defensible. Using it correctly is not a limitation. Using GT vocabulary incorrectly is a defense problem.