coding-pipeline

# Grounded Theory Coding Pipeline (Open → Selective → Theoretical) The coding pipeline is not a one-way assembly line. It is a **disciplined progression** where you repeatedly move between coding, comparing, sampling, and memoing until a theory **fits**, **works**, and earns **relevance**. ## Pipeline overview 1. **Open coding:** break data into incidents; label actions/meanings; keep comparisons tight. 2. **Selective coding:** elevate a core category; relate other categories to it; delimit the theory’s scope. 3. **Theoretical coding:** integrate categories using theoretical codes (families) to specify relationships among categories. Throughout: **memoing** captures hypotheses about relationships; **theoretical sampling** targets missing variation. ## Transition criteria (practical signals) Move toward **selective coding** when: - A category repeatedly explains variation and connects to many others. - Additional data mainly **densifies** rather than fractures the emerging core story. Move toward **theoretical coding** when: - Major categories are stabilized enough to ask **how they relate** (processually, conditionally, strategically). Return to open coding when **anomalies** appear—regression is healthy. ## Parallel processes Never pause memoing to “finish coding.” Memos are where theory grows. Schedule **memo-first** blocks after intense coding sessions. ## Quality checkpoints - **Fit check:** Do codes fit incidents without stretching? - **Work check:** Does the emerging theory explain how problems are handled/processes unfold? - **Relevance check:** Does it address the real concern in the data (not the literature’s concern)? - **Modifiability check:** Can you revise categories when new data demands it? ## Iteration patterns Common loops: - Open → memo → theoretical sampling → open. - Selective → theoretical → return to selective when integration breaks. Document each loop in an audit trail entry (“why we reopened category X”). ## Common stalling points - **Code sprawl:** too many codes at similar abstraction—schedule **code consolidation** sessions. - **Premature core category:** excitement about a pet theme—assign a **devil’s advocate** debriefer. - **Analysis paralysis:** set a rule—code N incidents, then write one **integrative memo**. ## Timeline expectations Timelines vary by data density and team size. A usable heuristic: expect **many cycles**; dissertations often spend months in open/selective interplay. Proposals should budget accordingly. ## Software note CAQDAS tools can help retrieval but cannot replace **constant comparison discipline**. Export periodic codebooks; version them. ## Handoff to writing When theoretical integration stabilizes, begin sorting memos into an outline. Writing is another pass of **theoretical refinement**. ## Checklist - [ ] Open coding grounded in incidents, not preconceptions. - [ ] Memos continuous; hypotheses dated. - [ ] Selective coding justified by centrality and recurrence in data. - [ ] Theoretical coding specifies relationships, not just labels. - [ ] Regression to earlier stages documented when anomalies arise. ## References (starting points) - Glaser, B. G. *Basics of Grounded Theory Analysis: Emergence vs. Forcing*. - Glaser, B. G. *Doing Grounded Theory: Issues and Discussions*. - Holton, J. A., & Walsh, I. *Classic Grounded Theory: Applications With Qualitative and Quantitative Data*.