Qualitative Research Methods: The Major Methodologies and How to Choose

“Qualitative research” names an approach — see what makes a study qualitative for the underlying definition. This guide covers the layer beneath that: the named methodologies researchers actually design studies within — phenomenology, grounded theory, ethnography, case study, narrative inquiry, and a few others — what philosophically distinguishes them from each other, and how to choose among them for a given research question. It does not cover the mechanics of coding and thematic analysis (see Thematic Analysis: A Step-by-Step Guide to Braun and Clarke’s Six Phases) or a head-to-head comparison with quantitative research (see Qualitative Research vs. Quantitative Research).

Methodology is not method

A “method” is a technique for collecting or analyzing data — an interview, a focus group, thematic coding. A “methodology” is the coherent design logic that determines which methods are appropriate, what counts as evidence, and how findings should be judged. This distinction matters because reviewers and methods-section critiques usually target it: a study that says “grounded theory” in its methods section but simply codes interviews for recurring themes without constant comparison, theoretical sampling, or a generated theory has borrowed the label without the design logic underneath it. This mismatch is sometimes called methodological slurring — combining incompatible traditions, or claiming one tradition’s name while actually practicing generic qualitative description. It is one of the most common reasons a qualitative methods section draws reviewer pushback, and it is avoidable by being explicit about which methodology is being used and following its actual design requirements, not just its vocabulary.

The major qualitative methodologies

Phenomenology: the structure of lived experience

Phenomenology asks what an experience is like for the people who have lived it, and aims to describe the essential structure shared across those individual experiences. It splits into two philosophically distinct schools that are frequently conflated:

• Descriptive (Husserlian) phenomenology asks the researcher to set aside — “bracket” — their own assumptions and prior knowledge about the phenomenon (a practice called epoché) in order to describe the experience as it presents itself, uncolored by the researcher’s interpretation.
• Interpretive/hermeneutic (Heideggerian) phenomenology rejects bracketing as neither fully achievable nor desirable: the researcher’s own background and interpretive lens are treated as an unavoidable and legitimate part of making meaning from the data, not a bias to eliminate. The aim shifts from pure description to interpretation of meaning within context.

A study that claims to “bracket” its findings while also foregrounding the researcher’s interpretive stance is mixing the two schools’ incompatible premises. For full treatment of both schools, plus Interpretative Phenomenological Analysis (IPA), sampling, and analysis procedure, see Phenomenological Research: Method, Approaches, and When to Use It.

Grounded theory: generating theory from data

Grounded theory does not test an existing theory against data — it generates a new, data-grounded theory of a social process through systematic, iterative analysis. It was introduced by Barney Glaser and Anselm Strauss in The Discovery of Grounded Theory (1967) as an alternative to the era’s dominant model of first deriving hypotheses from existing theory and only then collecting data to test them.

Three core mechanics run through every version of grounded theory:

• Constant comparison — each new piece of data is compared against previously coded data and emerging categories, continuously refining and testing those categories rather than coding once and stopping.
• Theoretical sampling — who or what to sample next is driven by what the emerging categories need (to test, extend, or challenge them), not by a sampling plan fixed in advance. Sampling continues until theoretical saturation: new data stops adding new properties to the core categories.
• Memoing — the researcher writes analytic memos throughout data collection and analysis, capturing the reasoning behind emerging categories; memos are themselves data for building the final theory, not just process notes.

Grounded theory subsequently split into three distinguishable traditions, and knowing which one a study follows changes what “doing it correctly” means:

• Glaserian (classic) grounded theory stays close to the 1967 original: theory should emerge from the data with minimal preconception, and Glaser argued researchers should defer an extensive literature review until after their own categories have emerged, specifically to avoid forcing data into pre-existing concepts.
• Straussian grounded theory, following Strauss and Corbin’s Basics of Qualitative Research (1990 and later editions), introduced a more structured, explicit coding sequence — open coding, axial coding (relating categories to one another through a conditions-actions-consequences framework), and selective coding (integrating categories around a core category). This is the version most commonly taught and used in applied and health-services research. Glaser publicly objected that this structure “forced” data into a preconceived framework rather than letting theory emerge, a disagreement that split the two originators’ approaches permanently.
• Constructivist grounded theory, developed by Kathy Charmaz (Constructing Grounded Theory, 2006), keeps the coding, memoing, and theoretical-sampling toolkit but rejects the idea of a neutral researcher discovering a theory that exists independently in the data. Instead, the researcher and participants are understood to co-construct meaning together, and the resulting theory is treated as one interpretation among possible others, not an objective discovery.

The literature-review-timing question above is a real, live methodological controversy, not a settled procedural detail — a methods section should state explicitly which grounded theory tradition it follows and be consistent with that tradition’s stance on it. See Grounded Theory for the concise definition and coding-terminology reference.

Ethnography: culture and social process from the inside

Ethnography studies a culture or social group’s shared patterns of behavior, language, and meaning through sustained, first-hand engagement with the setting — not a single interview or a short site visit. Its defining features:

• Prolonged engagement / participant observation — the researcher spends extended time within the setting, participating in its activities (to varying degrees, from full immersion to observer-as-participant) rather than only observing from outside.
• Fieldnotes — detailed, ongoing written records of what is observed, said, and done, typically distinguishing descriptive notes (what happened) from analytic/reflective notes (the researcher’s developing interpretation).
• Emic vs. etic perspective — the emic account describes meaning in the terms members of the culture themselves use; the etic account describes it in the researcher’s own analytic or theoretical terms. Rigorous ethnography holds both distinctly rather than collapsing one into the other.

Two variants adapt the tradition to different constraints. Autoethnography turns the method on the researcher’s own lived experience within a culture, using systematic self-reflection connected to broader cultural analysis, rather than an outside observer’s account. Focused (or rapid) ethnography compresses the fieldwork timeline — often to weeks rather than months or years — using more tightly scoped research questions, multiple data-collection methods run in parallel, and sometimes a research team rather than a lone fieldworker, to produce ethnographic insight under real time and resource constraints, common in applied and health-services settings. For the writing-side conventions specific to this tradition — thick description, reflexive first-person voice, how fieldnotes become a manuscript — see Ethnographic Writing Conventions: Thick Description, Reflexivity, and First-Person Fieldwork Voice.

Case study: a bounded system, not a small sample

A common misconception treats “case study” as simply a synonym for a small or single-participant sample. It is not — a case study is defined by its bounded system: a single case (a person, program, organization, event, or decision) or a small set of cases studied in depth, within its real-world context, using multiple sources of evidence. The boundary — what is inside the case and what is context around it — is a design decision the researcher must state explicitly, and it is what distinguishes case study from simply “a study with few participants.”

Two influential approaches diverge on epistemology and procedure:

• Robert Yin’s approach sits closer to a positivist logic: the researcher typically specifies propositions in advance, uses a formal case-study protocol, and — in multiple-case designs — applies replication logic, treating each additional case like a further experiment testing whether the same pattern holds (literal replication) or predictably varies (theoretical replication), rather than treating cases as a sample for statistical generalization.
• Robert Stake’s approach is more interpretive: it emphasizes understanding the particular case in its own right (intrinsic case study) or using a case to illuminate a broader issue (instrumental case study), with the researcher’s developing interpretation playing a more central, acknowledged role than in Yin’s protocol-driven design.

Designs are further classified as single-case vs. multiple-case and holistic vs. embedded (a holistic design treats the case as one unit of analysis; an embedded design examines multiple sub-units within the case, such as several departments within one organization). Case studies generalize analytically — extending a theoretical proposition to other contexts where similar conditions apply — rather than statistically to a defined population, which is the basis for the “small sample” misreading: case study isn’t attempting the kind of generalization sample size would fix. For the full design decision tree, worked examples of each type, and how case study differs from a clinical case report, see Case Study Research Method: Design, Types, and When to Use It, Case Study Examples, and Case Study vs. Case Report.

Narrative inquiry: stories as the unit of analysis

Narrative inquiry treats the stories people tell about their lives — not just the events in them, but how they are told, sequenced, and given meaning — as the primary data and object of analysis. It rests on the premise that people make sense of experience narratively, so the structure of a told story is itself analytically significant, not just a container for extractable facts.

• Life history / biographical narrative approaches collect an account of some or all of a person’s life, often through extended or repeated interviews, to understand how they construct identity and meaning over time.
• Restorying is a core analytic move: the researcher gathers a participant’s account, often non-chronological and fragmented as told, and reorganizes it into a coherent narrative sequence (typically with attention to setting, characters, conflict, and resolution), which is then usually returned to the participant to confirm it reflects their meaning.
• Structural narrative analysis examines how a story is told — its sequencing, linguistic devices, and narrative form — treating structure itself as data.
• Thematic narrative analysis instead focuses on what is told — the content and meaning of the story — while still preserving the story as a whole rather than fragmenting it into decontextualized codes the way generic thematic coding does.

Narrative inquiry differs from grounded theory and case study in its unit of analysis: it holds the story intact as the thing being analyzed, rather than breaking data into categories (grounded theory) or bounding a system (case study).

Briefly: action/participatory research and discourse analysis

Two further traditions are common enough to name here, with dedicated guides for full treatment:

• Action research is cyclical and interventionist: researchers and practitioners jointly plan, implement, and study a change to a real problem, then revise and repeat — the aim is change in the setting itself, not only an account of it. Participatory action research (PAR) and community-based participatory research (CBPR) push this further, treating the people affected by the problem as co-researchers with genuine authority over the research agenda, not just subjects or informants. See Action Research: Method, Cycle, and When to Use It, and the dictionary entries for action research, participatory research, and community-based participatory research (CBPR).
• Discourse analysis examines language in use — text and talk — as socially constitutive: how word choice, framing, and rhetorical structure produce and reproduce meaning, identity, and power relations, rather than treating language as a neutral window onto some separate reality it describes. See How to Conduct a Discourse Analysis.

How to choose: let the research question drive the methodology

The most common design error is picking a methodology by familiarity or convention rather than fit. The research question’s actual shape should point to the methodology, roughly along these lines:

• The question asks what an experience is like for those who lived it → phenomenology.
• The question asks how a process unfolds or seeks to explain/build theory about a social process with no adequate existing theory → grounded theory.
• The question is about a group’s shared culture, practices, or meaning-making over time → ethnography.
• The question concerns a bounded, contemporary phenomenon that can’t be separated from its real-world context (a program, an organization, a decision) → case study.
• The question is about identity, meaning-making over time, or how people story their own experience → narrative inquiry.
• The goal is to produce change in a real setting, with those affected as co-researchers → action/participatory research.
• The question concerns how language itself constructs meaning, identity, or power → discourse analysis.

Methodology choice should be finalized before data collection begins wherever possible — it determines the sampling logic, the data-collection method, and the rigor criteria the finished study will be judged against.

Paradigm: the assumptions underneath the methodology

Every methodology sits on a research paradigm — assumptions about the nature of reality (ontology) and how it can be known (epistemology). The major paradigms relevant to qualitative work are positivist/post-positivist (a single, knowable reality exists independently of the researcher — closest to Yin’s case-study approach and classic grounded theory), constructivist/interpretivist (reality and meaning are constructed through human interaction and interpretation — underlies hermeneutic phenomenology, Stake’s case study, and constructivist grounded theory), critical (research should expose and challenge power structures — underlies much action research and critical discourse analysis), and pragmatist (methodology should be chosen for what best answers the question, often combining approaches). A methods section should be internally coherent with its stated paradigm — claiming a constructivist stance while writing findings as objective, context-free facts is the same “methodological slurring” problem described above, applied at the paradigm level rather than the methodology level. For the full four-paradigm treatment, see Research Paradigm Explained: Positivism, Interpretivism, Pragmatism, and Critical Theory.

Sampling in qualitative research

Qualitative sampling is purposeful rather than random — participants or cases are selected because of what they can contribute to answering the question, not to represent a population proportionally. Common strategies:

• Purposive sampling — deliberately selecting participants who meet specific criteria relevant to the research question.
• Theoretical sampling — used specifically in grounded theory; sampling is directed by what the emerging theory needs next, not fixed in advance (see the grounded theory section above).
• Maximum variation sampling — deliberately selecting cases that differ widely on key characteristics, to capture the breadth of variation in an experience or process.
• Snowball sampling — existing participants refer additional participants, useful for hard-to-reach or hidden populations.

Sample size in qualitative research is governed by saturation — the point at which additional data collection stops producing new codes, themes, or theoretical properties — rather than a power calculation. A study should describe how and when it judged saturation reached, not just state a final sample size. See Snowball Sampling, Sampling Bias, and Purposive Sampling vs. Convenience Sampling for more.

Rigor: judged differently than quantitative validity

Qualitative research is not judged by the same reliability/validity criteria as quantitative work — those assume a single measurable reality independent of the researcher, which most qualitative paradigms explicitly reject. Instead, Lincoln and Guba’s widely used framework substitutes four parallel criteria:

• Credibility (parallel to internal validity) — confidence that the findings accurately represent participants’ meanings, often supported by prolonged engagement, triangulation, and member checking.
• Transferability (parallel to external validity) — whether findings could apply in other contexts, supported by thick, detailed description that lets a reader judge fit for their own setting, rather than a statistical generalization claim.
• Dependability (parallel to reliability) — whether the research process is logical, traceable, and documented well enough that another researcher could follow it, typically evidenced through an audit trail.
• Confirmability (parallel to objectivity) — whether findings are grounded in the data rather than the researcher’s own biases, supported by reflexivity and an audit trail connecting interpretations back to raw data.

Practical techniques supporting these criteria include reflexivity (the researcher explicitly examining and disclosing how their own position, assumptions, and relationship to participants shape the research, rather than presenting themselves as a neutral instrument), an audit trail (a documented record of analytic decisions from raw data to final themes/theory), member checking (returning findings or interpretations to participants to confirm they resonate), and triangulation (using multiple data sources, methods, or researchers to corroborate a finding). See Triangulation in Research for a full treatment; the quantitative analogues are covered in Construct Validity and Types of Validity in Research.

Reporting standards: COREQ and SRQR

Many journals now require a completed reporting checklist alongside a qualitative manuscript submission, and choosing the right one is itself a methodological decision worth getting right:

• COREQ (Consolidated Criteria for Reporting Qualitative Research) is a 32-item checklist specifically for studies using interviews or focus groups, and is the most commonly required checklist in health and clinical qualitative research.
• SRQR (Standards for Reporting Qualitative Research) is a 21-item checklist that applies more broadly across qualitative designs — including grounded theory, ethnography, phenomenology, and case study — not just interview/focus-group studies.

A related checklist, ENTREQ, applies specifically to qualitative evidence synthesis (meta-synthesis across multiple primary studies) rather than a primary study itself. All three are endorsed by the EQUATOR Network, the standard clearinghouse for health-research reporting guidelines. Checking a target journal’s author guidelines for which checklist it requires — and matching that checklist to the study’s actual design, not just defaulting to COREQ because it’s the most common — is a submission-readiness step research offices and journal-relations staff can usefully check for before manuscript submission.

Ethics specific to qualitative research

Qualitative designs raise ethical considerations beyond the standard informed-consent framework:

• Consent in emergent designs — ethnography, grounded theory, and action research often cannot fully specify in advance what will be observed, asked, or co-produced, since the design develops as data collection proceeds. Ethics review and informed consent processes for these designs typically rely on process consent (ongoing, re-confirmed as the study evolves) rather than a single fixed-scope consent obtained at enrollment.
• Confidentiality when quotes are identifiable — narrative, life-history, and case-study data are often detailed enough that even with names removed, a participant’s account, role, or circumstances may be identifiable to others in a small community, organization, or field. Anonymization needs to be judged at the level of the whole account, not just by stripping direct identifiers from individual quotes.
• Power asymmetry — the researcher-participant relationship in prolonged fieldwork, life-history interviewing, or organizational case study can create dependency or influence that a single-encounter survey rarely does; reflexivity about this relationship is itself part of rigorous reporting, not just an ethics-review formality.
• Secondary use of narrative and fieldwork data — because qualitative data is rich, contextual, and often identifiable even after de-identification, secondary use or archiving typically requires more careful review than quantitative data, and participants’ original consent scope needs to be checked before data is reused or shared for a different purpose.

Frequently asked questions

What is a qualitative researcher?

A qualitative researcher is someone who designs and conducts studies using one or more of the qualitative methodologies above — selecting methodology to fit a meaning- or process-focused research question, collecting non-numeric data (interviews, observation, documents, artifacts), and analyzing it inductively rather than testing a pre-specified hypothesis. See Qualitative Research for the full operational definition.

Is qualitative research less rigorous than quantitative research?

No — it is rigorous by different, paradigm-appropriate criteria. Lincoln and Guba’s credibility/transferability/dependability/confirmability framework (above) is the qualitative equivalent of quantitative validity and reliability, not a lesser substitute for it. A qualitative study that skips reflexivity, an audit trail, or an appropriate sampling logic for its methodology is a poorly executed qualitative study — the same way an underpowered survey is a poorly executed quantitative one — not evidence that the qualitative approach itself is inherently weaker.

Can a single study combine more than one qualitative methodology?

It can, but doing so deliberately and coherently is harder than it looks — each methodology carries its own paradigm, sampling logic, and analytic procedure, and combining them without addressing that (for example, borrowing grounded theory’s coding vocabulary while doing an ethnography) is exactly the “methodological slurring” problem described above. Studies that genuinely need more than one tradition should state explicitly how the traditions are being combined and why each is warranted by a distinct part of the research question.

How is qualitative research different from mixed-methods research?

Qualitative research uses non-numeric data and inductive analysis throughout. Mixed-methods research deliberately combines qualitative and quantitative strands within a single study design (for example, a quantitative survey followed by qualitative interviews to explain an unexpected result), with its own set of design frameworks distinct from any single qualitative methodology.