Glaserian Grounded Theory

When in doubt, privilege: emergence, comparison, memoing, and theoretical sampling directed by the emerging theory.

The complete GT process (ten integrated stages)

These stages are iterative, not strictly linear. You cycle among them throughout a study.

1. Preparation and entry
   Clarify a broad substantive area of interest (not a forced research question). Secure ethics permissions. Set up data management and audit trails.

2. Data collection (initial)
   Begin with purposive (not theoretically sampled) first interviews/observations to get rich incidents. Treat everything as potential data.

3. Open coding
   Break data into incidents and label them with substantive codes. Compare incident to incident. Write memos capturing ideas.

4. Constant comparison
   Continuously compare incidents to incidents, incidents to concepts, and concepts to concepts. Refine code definitions, properties, and dimensions.

5. Memoing
   Never treat memoing as optional. Memos are the record of conceptual leaps—hypotheses about relationships, conditions, consequences, and processes.

6. Selective coding
   Once a core category emerges and earns centrality, delimit the study: code only for what relates to the core and its story.

7. Theoretical coding
   Relate categories using theoretical codes (coding families) to build an integrated theoretical outline.

8. Theoretical sampling
   Collect new data to develop emerging categories—not for representativeness alone. Sample for variation, depth, and theoretical completeness.

9. Theoretical saturation
   Stop sampling for a category when new data no longer yield new properties or relationships relevant to the emerging theory.

10. Theory write-up
    Present a substantive theory: core category, related categories, and theoretical statements (hypotheses) about relationships, often centered on a basic social process or core pattern.

Key principles (non-negotiable in classic GT)

Emergence

Categories and hypotheses must earn their way from comparative analysis. If a code “doesn’t work,” discard or modify it. Forcing pre-existing frameworks onto data violates the logic of the method.

All is data

Interviews, observations, documents, artifacts, field notes, researcher reflections, and later literature (introduced at the right time) can be treated as data. Nothing is a priori “off-limits,” though ethics and scope still bound what you should collect.

Constant comparison

Comparison is the engine of analysis. If you are not comparing, you are likely drifting into description or affirmation of prior beliefs.

No preconception (disciplined openness)

Enter the substantive area without a pre-rehearsed conceptual framework. Theoretical sensitivity is cultivated through analytic work and broad reading outside the substantive area—not by importing a ready-made model of the phenomenon.

Fit, work, relevance, modifiability

Glaser’s theory criteria (see substantive-theory skill):

• Fit: categories fit the data used to generate them.
• Work: theory explains variation and the main pattern.
• Relevance: addresses what participants actually resolve (their “main concern”).
• Modifiability: open to refinement with new data—not brittle dogma.

Coding procedures in depth

Open coding

• Fracture narratives into discrete incidents (meaningful chunks).
• Ask: “What is this data a study of?” “What category does this incident indicate?”
• Use gerunds when helpful to keep process visible (e.g., managing, balancing, negotiating).
• Generate in vivo codes (participants’ language) when they crystallize meaning.
• Write memos immediately when ideas spark.

Selective coding

• Identify the core category: recurrent, central, relates to most other categories, explains variation.
• Delimit: stop coding everything; focus on the story that integrates the core with related categories.
• Aim for theoretical completeness around the core, not encyclopedic coverage.

Theoretical coding

• Use theoretical codes (e.g., causal, conditional, strategy) as integrative devices—not as a front-loaded template.
• Apply coding families thoughtfully (see theoretical-coding skill).
• Build a theoretical outline that states relationships among categories.

Memoing as the core intellectual activity

Memos are analytic narratives: they explain what you think is going on conceptually, propose relationships, note puzzles, and log methodological decisions.

Rules of thumb:

• Stop coding when a big idea hits—memo first, then return.
• Sort memos regularly into an emerging outline of the theory.
• Date memos and link them to data locations (interview line, page, timestamp).
• Allow memos to contradict earlier memos; treat contradictions as opportunities for comparison.

Theoretical sampling (brief)

Sampling is driven by analysis, not only by access or convenience. You ask: “What data do I next need to develop this category’s properties, dimensions, and relationships?”

See theoretical-sampling skill for directives, probes, and pitfalls.

Theoretical saturation (brief)

Saturation is about categories, not raw repetition of stories. A category is saturated when continued sampling does not refine its properties, dimensions, or relationships in ways that matter to the emerging theory.

See theoretical-saturation skill for checklists and common confusions.

Writing the theory

A classic GT write-up foregrounds:

• The main concern and how participants continually resolve it.
• The core category and its multivariate relations (conditions, strategies, consequences).
• A processual account if a basic social process (BSP) is core.
• Conceptual language with illustrative (not exhaustive) data excerpts.
• Theoretical statements that read as hypotheses grounded in comparative evidence.

Avoid “theme lists” without integration. Avoid over-quotes with thin conceptual lift.

Common mistakes and how to avoid them

Comparison with other GT traditions (practical distinctions)

• Classic Glaserian: emergence-first; theoretical codes used after substantive theory forms; strong stance against forced paradigms.
• Strauss & Corbin style: often teaches structured coding steps (e.g., axial) and conditional matrix tools earlier and more prescriptively—useful for some teams, but not identical to Glaser’s classic GT.
• Constructivist GT (Charmaz): shares inductive spirit; emphasizes interpretive construction and researcher-participant meaning-making—aligns ethically/politically with many projects; epistemological language differs from Glaser’s objectivist-realist emphasis.

If your project must integrate another tradition, document that choice in your audit trail and clarify what you borrowed and why.

Key references

• Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for qualitative research. Aldine.
• Glaser, B. G. (1978). Theoretical sensitivity: Advances in the methodology of grounded theory. Sociology Press.
• Glaser, B. G. (1992). Basics of grounded theory analysis: Emergence vs forcing. Sociology Press.
• Glaser, B. G. (1998). Doing grounded theory: Issues and discussions. Sociology Press.
• Glaser, B. G. (2005). The grounded theory perspective III: Theoretical coding. Sociology Press.

Companion skills in this library

• open-coding, selective-coding, theoretical-coding
• constant-comparison, memo-writing
• theoretical-sampling, theoretical-saturation, theoretical-sensitivity
• substantive-theory, formal-theory

Quick-start checklist (classic GT)

• Broad area of interest defined; ethics in place
• First data collected and fractured into incidents
• Open coding + incident comparison running in parallel with memoing
• Codesheet/evolving codebook + audit trail of renames/splits/merges
• Core category candidate tested against fit and explanatory reach
• Selective coding + theoretical coding toward an integrated outline
• Theoretical sampling to saturate key categories
• Draft theory written in conceptual voice with clear hypothetical statements

Notes on AI-assisted analysis

AI tools can assist fracturing, draft coding, memo prompts, and outline experiments—but the researcher must own comparisons, verify fit against source data, and maintain an audit trail. Treat AI outputs as provisional hooks for comparison, not as findings.