Stages in Developing Grounded Theory

Grounded theory analysis works through three coding stages. Each stage handles a different level of abstraction, and the move from one to the next is what builds theory from data. The stages are not rigid steps; they overlap and feed back into each other. But the sequence matters.

Three coding stages

The three stages are open coding, axial coding, and selective coding.

Open coding: developing categories

Open coding is the first stage. The researcher reads through the data and applies codes that label what they see. The goal is to develop categories of information from the data.

The work is descriptive at this stage, but the descriptions are deliberate. A useful discipline is to avoid mere description. Instead of writing “talked to a manager”, the researcher labels the action as “conferring”. Instead of “reading the schedule”, they label it as “information gathering”. The codes are at a slightly higher level than the surface description.

Open coding uses the constant comparative approach. Each new piece of data is compared with what has already been coded. New codes are created when the data does not fit existing ones. Existing codes are refined when new data adds nuance.

The researcher continues open coding until saturation. Saturation means looking for instances that represent the category and finding that new instances do not provide further insight. The category is full.

The discipline during open coding is to focus on action. The researcher asks: what is the person doing? What are they trying to achieve? What strategy are they using? Coding social and psychological processes keeps the analysis aligned with grounded theory’s interests.

The process is iterative. Coding builds up gradually, with later codes informed by what was learned in earlier ones. The first pass through the data produces rough codes. The second pass refines them. The third pass refines further.

Axial coding: connecting categories

Axial coding is the second stage. The researcher explores the relationships between the categories that emerged from open coding. The categories are no longer treated as separate; the researcher looks for how they connect.

A common model used in axial coding, drawn from Strauss and Corbin, has six elements arranged in a sequence.

1. Causal conditions. What influences the central phenomenon? Events, incidences, happenings that bring it about.
2. Central phenomenon. The central idea, event, or happening that the analysis is trying to understand. The thing being studied.
3. Context. The location and situation in which the phenomenon happens.
4. Intervening conditions. Conditions that affect how the phenomenon plays out, modifying its course.
5. Action and interaction strategies. What people do in response to the phenomenon. Purposeful, goal-oriented actions.
6. Consequences. What follows from the actions and interactions. The outcomes.

A teacher analysing student responses to feedback might use this model. The central phenomenon is the response to feedback. Causal conditions include the kind of feedback given, the relationship between teacher and student, the student’s prior experience. Context is the classroom and the moment. Intervening conditions include the student’s emotional state, the time of day, what else is happening in the lesson. Action and interaction strategies are what the student actually does in response. Consequences are what follows.

The axial coding model gives the analysis a structure that pushes it past descriptive accounts and toward an analytic account of how the phenomenon works.

Selective coding: building a story

Selective coding is the third stage. The researcher identifies a single category as the central phenomenon and constructs a story around it.

The story line is the conceptualisation of the situation. It is the core category that ties the rest of the analysis together. Selective coding systematically relates the core category to other categories and fills in any categories that need further refinement.

The theory at this stage emerges by constant comparison; it is not forced. The phrase emergent matters. The researcher is not deciding what the story is; they are finding the story that the data supports.

The output of selective coding is the theory. It is the answer to the research question, expressed as an account of how the central phenomenon works in this setting.

A useful test of a selective coding result is whether the theory is prescriptive about its own development. The theory should specify the categories it depends on, how they relate, and what conditions modify their effects. A grounded theory that is too vague to test against new data has not been developed enough.

Coding gets to the concept, not the description

A central discipline across all three stages is that coding aims at concept, not description. This is worth dwelling on.

Description tells you what happened. “The student raised their hand and asked a clarifying question.” A code that captures the description is at the surface.

Concept tells you what the action is an example of. “Engaged help-seeking.” This concept can apply to many specific actions and can be related to other concepts.

Grounded theory studies a concept, not the surface events. The same teacher’s classroom might show many surface events that all serve the concept of “negotiating ambiguity”. A grounded theory study about that classroom would build a theory about negotiating ambiguity, with the surface events as evidence, not as the theory itself.

This shift from description to concept is what makes the difference between description and theory. Without it, the study produces a chronicle. With it, the study produces an account of how something works.

Theoretical sensitivity

Theoretical sensitivity is the researcher’s ability to see what is theoretically interesting in the data. It comes from several sources.

Professional experience

A teacher who has spent years in classrooms has a stock of professional experience that helps them see what matters in a teaching situation. They know what is unusual and what is routine. This experience is a source of theoretical sensitivity.

Personal experience

The researcher’s own life experiences can also help. A teacher who has experienced both private and public education will have a theoretical sensitivity to social class differences in classrooms that another teacher might lack.

The analysis process itself

The third source is the analysis process itself. As the researcher works through the data, they develop sensitivity to concepts, meanings, and relationships in this specific setting. The analysis trains the analyst.

Theoretical sensitivity is a useful resource and a possible distortion. It helps the researcher connect what they know to what they are studying. It also raises the risk that the researcher’s biases shape what they see.

Theoretical sensitivity therefore acts as a grounding tool. It forces the researcher to step back and check that what they think they are seeing is actually found in the data, not projected onto it. The phrase grounded in grounded theory points at this discipline.

The interplay of philosophy, method, and data

The full picture of grounded theory work is a continuous interplay across several levels.

At the bottom level are the philosophical assumptions of the researcher. What do they believe knowledge is? What counts as evidence?

Above this is the research method itself. Grounded theory has its assumptions and procedures, drawn from the philosophical level.

Above the method is the data collection technique. Interviews, observations, documents, video, the choice of technique flows from the method.

Above the technique is the data analysis approach. Coding stages, constant comparison, theoretical sampling.

At the top is the written record: the report, paper, or thesis that comes out of the work.

Each level shapes the level above it. The philosophical assumptions shape the method. The method shapes the technique. The technique shapes the analysis. The analysis shapes the written record. A grounded theory study is coherent when these levels all line up.

Flashcard

What are the three stages of coding in grounded theory?

Tap to reveal

Answer

Open, axial, selective

Open coding develops categories from the data through descriptive labels and constant comparison. Axial coding connects the categories through a model of causes, contexts, intervening conditions, strategies, and consequences. Selective coding identifies a core category and builds the theory as a story around it.

How a teacher might use the stages

A teacher running a small grounded theory study in their classroom would work through the three stages.

First, they collect data. This might be lesson observations they record, student work samples, audio of class discussions, written reflections, or interviews with a few students.

Then they begin open coding. They read the data carefully, label what they see, and develop initial categories. They compare each new piece of data to what they have already coded. They continue until the categories feel saturated.

Next, they move to axial coding. They look at how the categories relate. They apply a model like Strauss and Corbin’s to map the relationships. The map starts to look like a small theory.

Finally, they move to selective coding. They pick the central phenomenon their study is about and write the story that ties the categories together. The story is the theory.

The whole process is slower than informal reflection but produces a more disciplined understanding. A teacher who has done one or two small grounded theory studies usually returns to ordinary reflection with sharper habits of attention.