What Is Axial Coding? A Practical Guide

Axial coding is the process of relating codes and categories to each other. You take the pieces open coding produced and connect them around a phenomenon: its conditions, the actions or interactions it triggers, and its consequences. It's stage two of grounded theory's three-stage sequence: open coding, axial coding, selective coding.

If you've just finished open coding a batch of interviews and you're staring at forty loose codes with no idea how they relate to each other, axial coding is the step that turns that pile into structure.

Where does the term come from?

Axial coding was formalized by Anselm Strauss and Juliet Corbin in their 1990 and 1998 textbooks on grounded theory methodology. They proposed a "coding paradigm," a model for organizing categories around four elements:

• The phenomenon – the central idea or event the data is about
• Conditions – the circumstances that give rise to it
• Actions/interactions – how people respond to or manage it
• Consequences – the outcomes of those actions

This paradigm gave researchers a repeatable way to ask, of every category: what causes this, what happens because of it, and how do people act around it. It's a deliberately structured approach, rooted in pragmatist and interactionist social theory. Strauss and Corbin wanted grounded theory to produce not just a list of themes, but an explanatory model of action.

How it fits into the coding sequence

1. Open coding – you break the data into discrete pieces and label them, staying close to what's actually said
2. Axial coding – you put the pieces back together in a new way, by identifying relationships between codes and grouping them into categories and subcategories
3. Selective coding – you integrate and refine the categories around a core category, building the final theoretical structure

Axial coding bridges the gap between raw fragments and an actual analytic structure. Skip it, and open coding just leaves you with a long, flat list.

A short worked example

Say you're coding interviews with office employees about using AI tools at work. Open coding gives you fragments like "I don't tell my manager which parts ChatGPT wrote," "everyone's suddenly twice as fast and nobody explains how," "I feel like I'm cheating even when the work is good." Axial coding asks: what connects these? You might land on a category like quiet competence anxiety, with conditions (rising output expectations, no formal guidance on tool use), actions (using AI privately, hiding the assist from colleagues), and consequences (eroded trust in your own skills, reluctance to ask for help). That category, not the individual quotes, is what goes into your findings.

Axial coding vs. open coding

The short version: open coding breaks data apart, axial coding puts it back together around relationships. See our companion guide, Open Coding vs. Axial Coding: What's the Difference, for a side-by-side comparison.

Is axial coding always necessary?

Not everyone agrees it should be. Kathy Charmaz, a leading figure in constructivist grounded theory, calls axial coding "highly structured and optional": useful for some studies, not obligatory for all of them. If your categories already show clear relationships after open coding, or your epistemological stance leans more interpretive than the original Straussian model, skip the formal pass. Nobody hands out points for extra steps. The point of the method is analytic clarity, not procedural compliance, and that's a caution echoed more broadly in recent methodological literature: mechanically applying a fixed step (sample saturation is the usual example) without regard to a method's own epistemological logic tends to flatten the analysis rather than strengthen it.

How this looks in practice with AI assistance

Be skeptical of any tool that claims to "automate" axial coding. The relationships between categories are an interpretive judgment call, not a pattern-matching task. What AI can reasonably do is suggest first-pass groupings of related codes for a researcher to review, rename, merge, or reject. That's the kind of human-in-the-loop workflow Qualibr is built around: AI proposes structure, but every category, relationship, and citation stays editable and traceable back to source material. That distinction – AI as a first draft, not a verdict – is worth applying whatever software you use.

FAQ

What's the difference between axial coding and thematic analysis? They come from different traditions. Axial coding is a specific step within grounded theory, built around Strauss and Corbin's coding paradigm (conditions, actions, consequences). Thematic analysis, especially Braun and Clarke's approach, is a more flexible method for identifying patterns of meaning across a dataset, without the grounded-theory coding paradigm. See our guide to Braun & Clarke's six-phase thematic analysis framework.

Do I need special software to do axial coding? No. Axial coding is a way of thinking about your codes, not a piece of software. People do it with sticky notes, spreadsheets, or dedicated QDA tools. Software mainly helps by making it faster to see which codes co-occur, and by keeping the relationships traceable back to your source quotes.

Is axial coding the same as creating a codebook? Related but not identical. A codebook documents your codes and their definitions. Axial coding is the analytical work of deciding how those codes relate to each other. The output of axial coding often becomes the structure your codebook's categories are organized around.

Sources

• Strauss, A. & Corbin, J. (1990, 1998). Basics of Qualitative Research: Grounded Theory Procedures and Techniques.
• Axial Coding – The SAGE Encyclopedia of Qualitative Research Methods (reference entry on Strauss & Corbin's coding paradigm)
• Charmaz, K. (2006). Constructing Grounded Theory.
• Lyhne, I. & Norlyk, A. (2026). "Saturation – a one-size-fits-all fallacy." Qualitative Research. (used for the analogous methodological point about mechanical criteria)