Grounded theory in qualitative research: a complete guide

Summary:

Most research begins with a theory and tests it. Grounded theory begins with data and builds from there.

The difference sounds like a procedural detail. It isn't. The choice to start without a hypothesis, to let explanations emerge from evidence rather than confirm or deny what you already believe, is a fundamental methodological commitment.

It shapes how you sample, how you analyze, what you produce, and what kind of knowledge claim you can make at the end.

This guide covers what grounded theory is, how it differs from related concepts it's often confused with, and what the actual process looks like when you put it into practice.

Grounded theory is a qualitative research methodology that generates theory by systematically collecting and analyzing data, letting explanations emerge from evidence rather than testing a predetermined hypothesis.

The method was developed by sociologists Barney Glaser and Anselm Strauss and published in "The Discovery of Grounded Theory" in 1967. Their argument was both straightforward and, at the time, provocative: instead of testing existing theory against data, researchers should let theory grow from data. You begin without a hypothesis. You follow the evidence until an explanation takes shape on its own terms.

Since that original formulation, three distinct strands of grounded theory have developed:

The core commitment across all three strands is the same: theory is built from data, not imposed on it. The researcher does not arrive with conclusions to verify.

What grounded theory is not is worth stating directly, because the term gets misapplied:

Exploratory research is a broad category of early-stage research aimed at understanding a phenomenon before hypothesis testing. It includes:

All grounded theory is exploratory, but most exploratory research is not grounded theory. The distinction matters: grounded theory has specific, disciplined procedures that most exploratory research does not follow. Calling a study "grounded theory" because it's exploratory and qualitative is not accurate.

Qualitative research encompasses ethnography, case study, phenomenology, content analysis, narrative inquiry, discourse analysis, and many other methods.

Grounded theory is one specific methodology within that broader category, with its own logic, procedures, and epistemological commitments.

This is a common misunderstanding, particularly among researchers new to qualitative methods.

Grounded theory has specific, demanding procedures: iterative coding, constant comparative analysis, theoretical sampling, memo writing, and a defined stopping criterion in theoretical saturation. It is among the more procedurally rigorous qualitative methodologies.

Studies that describe themselves as grounded theory but omit systematic coding, constant comparison, or theoretical sampling are typically applying the label incorrectly.

Grounded theory matters most when existing theory is absent or inadequate, and you need to understand a phenomenon on its own terms.

Four contexts illustrate where the method earns its keep:

When a domain is too new for established frameworks, grounded theory builds the framework.

Researchers studying the social dynamics of early social media, the lived experience of remote work before it became widespread, or patient experience with a newly available treatment had no prior theory to test. They needed to generate explanations from scratch.

Grounded theory is designed precisely for that situation, where the existing literature doesn't yet have the vocabulary to name what you're observing.

Most quantitative research tests existing theory. It asks: does X predict Y, given this hypothesis? Grounded theory builds new theory. It asks: what is happening here, and what explains it?

These are different kinds of contributions, and both are necessary for a research field to advance. Grounded theory fills the front of the research pipeline, generating explanations that later researchers can formalize and test at scale.

A field that only tests theory and never builds it eventually runs out of things worth testing.

Product teams and UX researchers use grounded theory methods to understand user behavior when existing categories don't fit.

Why do users adopt a feature in ways the product team never anticipated? Why does one customer segment churn before completing onboarding while another doesn't? What does "trust" actually mean to someone considering a new financial product?

These questions don't have established theory to test. They require inductive investigation. Grounded theory gives that investigation structure and rigor.

Grounded theory is a standard methodology in UX research, organizational behavior, healthcare research, social work, and education.

If you encounter a published study in any of those fields describing iterative sampling, constant comparison, and theoretical saturation, you're reading a grounded theory study.

Familiarity with the methodology helps you evaluate the quality of published research and design your own studies to comparable standards.

Each new piece of data—an interview excerpt, a quote, an observation note—is compared with existing categories as soon as it's collected. That comparison refines and differentiates categories continuously throughout the study. Unlike research methods that separate data collection from analysis into sequential phases, grounded theory interleaves them. You code as you collect, and each round of analysis shapes what you collect next.

Constant comparison is not a step; it's a practice. You're asking of every new data unit: what category does this belong to? How is it similar to what I've already coded? How is it different? Does this confirm a category or challenge one? The answers to those questions direct the study forward.

Grounded theory's sampling logic is fundamentally different from predetermined sampling designs. In a standard survey study, you decide at the outset how many participants you need and what their characteristics should be. In grounded theory, you let your emerging theory direct sampling decisions. When your categories point to a gap, you recruit participants who can fill it. When your theory implies a boundary condition, you seek participants who test that boundary. You follow the theory to its edges.

This means you cannot fully plan your sample before a grounded theory study begins. You can plan the first phase. The rest of the sampling is determined by what you find. This is theoretically sound but practically challenging for timelines and budgets, which is worth accounting for when you scope a project.

Grounded theory coding typically moves through three stages:

Open coding is the initial pass. You read transcripts or field notes line by line and assign labels to the concepts you find. The goal is to break data into its component parts and name them. Glaser and Strauss encourage using participants' own language when possible — these are called "in vivo codes." Using participants' language preserves the meaning they attached to their experience and prevents the researcher's vocabulary from overwriting it prematurely.

Axial coding groups the concepts identified in open coding into categories, then identifies the properties and dimensions of each category. A category might be "managing uncertainty." Its properties might include source of uncertainty, strategies used, and emotional response. Axial coding builds structure and relationship between concepts.

Selective coding identifies the core category: the central concept that integrates all others. At this stage, you write the theoretical narrative around the core category, articulating how the other categories relate to it and to each other. This is where the study's contribution takes final shape.

Researchers write analytical memos throughout the study, from the first interview through the final theory. Memos are not just notes on what happened. They are the interpretive work of theory development: why two categories seem related, what a new interview revealed about a category you thought was settled, how a concept is shifting across the data, where the theory has gaps.

Memos bridge raw data and finished theory by documenting the researcher's thinking as it develops. They also serve a practical purpose: when you sit down to write the final theory, the memos give you a record of how you got there. Experienced grounded theory researchers often say the memos are as important as the interview transcripts.

You stop collecting data when new data confirms your theory but no longer extends it. That's the criterion of theoretical saturation, and it's the grounded theory stopping rule that most clearly distinguishes the method from research designs with fixed sample sizes.

Saturation cannot be predetermined; you recognize it when you reach it. You've interviewed five more people and none of them introduced a new category, modified an existing one, or challenged your theoretical model. The categories are stable. The relationships between them are clear. The theory accounts for the variation in your data. At that point, additional data collection adds confirmation but not insight.

In practice, saturation typically occurs somewhere between 20 and 40 interviews in published studies, but the range is wide. Some phenomena saturate earlier; others require more. The criterion is saturation, not the number.

Grounded theory research follows an iterative cycle of data collection, coding, and comparison rather than a linear sequence of steps.

That iterative quality is not a stylistic preference; it's built into the method's logic. You cannot collect all your data and then analyze it, because what you find in early analysis shapes what you collect next. Plan for collection and analysis to run in parallel from the beginning.

Your research question should open the inquiry rather than constrain it. "What is happening for people when they navigate this process?" works. "Does X cause Y?" does not. The question should invite description and explanation, not confirmation or denial of a prediction. The question will often sharpen as the study progresses, but it should start broad enough that it doesn't presuppose its answers.

Five to ten interviews is a common starting point, supplemented in some studies by observation notes, documents, or secondary sources. The goal at this stage is to generate enough material to begin coding and to surface the initial concepts that will anchor your theoretical sampling decisions.

Don't wait to analyze. After each interview or observation session, read the transcript line by line and code it before the next collection event. Name concepts as specifically as you can. Use participants' own language. This is time-consuming, but the conceptual work you do in early coding shapes the questions you ask in later interviews, and that shaping is how grounded theory works.

Document your analytical thinking as it develops. Why do these two concepts seem related? Where is this category unclear? What do you need to find out next? Memos capture the interpretive work that would otherwise happen only in your head. They also create a record that makes the final theory-writing phase significantly more tractable.

As you move through the data, compare each new piece against your existing categories. What fits? What challenges? What differentiates? Note when new data confirms a category, when it suggests a new one, and when it challenges an assumption you've been carrying. Constant comparison is not a phase; it runs throughout the study.

After your initial coding, identify what your emerging theory needs to develop. Which categories are underdeveloped? What boundary conditions haven't you explored? What kinds of participants or data sources would help you test the theory you're building? Recruit or seek those specifically. This is what makes grounded theory's sampling logic different from a standard qualitative study design.

Keep collecting, coding, comparing, and memo-writing until new data adds nothing new. That's saturation. You can't set a calendar date for it in advance; you recognize it when your categories are stable and your theory accounts for the variation in your data without generating new questions.

Identify the core category, the central concept that integrates all others, and write the theoretical narrative around it. Articulate how the other categories relate to the core and to each other. The output of a grounded theory study is not a list of themes; it's an explanatory model. It proposes not just what happened, but why, and under what conditions.

Open-ended surveys can generate initial data for grounded theory coding, particularly when interviewing every participant at scale is impractical.

In-depth interviews remain the most common primary data source in grounded theory research.

An interview allows the researcher to probe, follow unexpected threads, ask participants to elaborate on their own language, and redirect based on what emerges.

That responsiveness is central to grounded theory's iterative logic. Interviews allow the kind of emergent, theoretically directed data collection that theoretical sampling requires.

Open-ended survey questions produce a different kind of data: wider in scope, less deep, and less flexible. But they have a legitimate role at the opening of a study.

If you're researching how employees experience a new performance management system, an open-ended survey sent to 200 people can surface the range of concepts and concerns before you design the interviews that develop your theory.

The breadth of survey data gives you an initial conceptual map; the depth of interviews fills it in.

SurveyMonkey supports open-ended question types that generate transcript-like text suitable for initial open coding. The survey templates library and market research resources can help you design the initial data collection phase.

To be clear about where surveys fit: they are most useful at the data collection phase of grounded theory, specifically for generating initial categories from a larger population than interviews can reach.

The analysis phases—coding, constant comparison, memo writing, theory development—require qualitative analysis work that a survey platform does not provide.

Use surveys to cast a wide initial net. Use in-depth interviews and systematic coding to develop the theory that net reveals.