selective-coding

# Selective Coding (Classic Grounded Theory) Selective coding begins when a **core category** has **emerged** and **earned** its place as the center of the developing theory. You shift from coding **everything** to coding **only** what develops the **story** that integrates the core category with major related categories (conditions, strategies, consequences, etc.). Use this skill when your open coding and constant comparison suggest a **central pattern**, and you need to **delimit** scope while increasing **theoretical density** and **integration**. --- ## What selective coding is (and is not) **It is**: - **Focused** coding around the **core** and its multivariate relations. - A disciplined way to **stop sprawl** while deepening **explanatory** power. - A bridge from **many codes** to a **coherent theoretical outline**. **It is not**: - Cherry-picking quotes to fit a pet theory. - Ignoring contradictory data (negative cases remain essential). - Replacing comparison with a **single storyline** that ignores variation. --- ## Identifying the core category Glaser describes the core as the category that: - **Recurs** frequently in varied forms. - **Relates meaningfully** to many other categories. - Helps **explain** variation (why some incidents differ). - Often connects to participants’ **main concern** and how they **continually resolve** it. ### Practical tests for a core candidate Ask: 1. **Coverage test**: If I remove this category, does the theory **fall apart**? 2. **Connection test**: Does it **link** multiple other categories (not just sit beside them)? 3. **Process test** (when relevant): Does it read as a **basic social process** (BSP) or central social-psychological process? 4. **Variation test**: Does it help explain **differences** across incidents and cases? If a candidate fails these tests, keep comparing—**do not promote** a core by fiat. --- ## Basic social process (BSP) In many Glaserian studies, the core is a **process**: something participants **do** or **work through** over time (e.g., *status passage*, *becoming*, *managing uncertainty*). **Indicators you may be dealing with a BSP core**: - Temporal language (*then*, *after*, *at first*, *eventually*). - Stages/phases that **vary** by condition but retain a **family resemblance**. - Strategies that appear as **recurrent tactics** tied to conditions/consequences. Not every core must be a BSP, but if your data are strongly processual, **gerund formulations** often help. --- ## Delimiting the theory **Delimiting** means you intentionally **stop** trying to code all possible interesting topics. Instead, you ask of each new incident: > “What does this contribute to **the core** and its **integrated** relationships?” ### Delimiting rules of thumb - **Retain** data that clarifies **properties/dimensions** of core-related categories. - **Retain** negative cases that define **boundaries** and **conditions**. - **Deprioritize** tangents that do not change your **theoretical statements** even if they are “interesting.” Document delimiting decisions in **memos** to preserve transparency. --- ## Focused coding procedure 1. **State the core category** in a memo in **one sentence** (provisional). 2. List **major related categories** already suggested by comparison (conditions, context features, strategies, outcomes). 3. For each new data segment, ask: - “How does this **develop** the core story?” - “Does it add **properties**, **dimensions**, **conditions**, or **consequences**?” 4. **Compare** new incidents to **theoretical statements** you are beginning to trust—then revise statements if mismatch appears. 5. **Sort memos** into an emerging **outline** structured around the core. --- ## Transitioning from open to selective coding ### Signals you may be ready - Repeated **redundancy** in open coding relative to the same conceptual points. - Memo outlines keep returning to the **same center of gravity**. - You can **summarize** the study’s main pattern without relying on dozens of disconnected themes. ### A safe transition practice Run a **two-track week**: - Track A: continue open coding on **one** new transcript. - Track B: selective-code **only** for core development on the same transcript. Compare results: does selective coding **lose** essential variation? If yes, you’re not ready or your core is wrong. --- ## Relationship to theoretical coding Selective coding clarifies **what** must be integrated. **Theoretical coding** (coding families) clarifies **how** categories relate (e.g., conditional, causal, strategy). Sequence (typical in classic GT): 1. Substantive coding matures (open → selective). 2. Theoretical codes help **write** integrated hypotheses as the outline stabilizes. Avoid bolting on theoretical codes **before** substantive relationships are comparatively earned. --- ## Output format (recommended) ### Core statement template ```text Provisional core category: One-sentence core claim (hypothesis): Main concern (if identifiable): Primary related categories (list): Negative cases queued for comparison: What would falsify this core claim in upcoming data? ``` ### Selective coding excerpt log ```text Source ID / location: Why this excerpt matters to the core: New properties/dimensions discovered (if any): Revised relationship statements (if any): Next sampling impulse (theoretical sampling note): ``` --- ## Common mistakes - **Premature selective coding** after one interview. - **Renaming** without comparative justification (cosmetic “core”). - **Suppressing outliers** that define scope conditions. - **Theme rebranding**: calling a theme a “core” without **explanatory** work. - **Mixing units**: confusing participant goals with analytic categories—keep analytic language in memos and theory. --- ## Quality criteria checkpoint Revisit **fit, work, relevance, modifiability**: - **Fit**: Does the core fit **all** major data contexts you’ve compared (including deviant cases)? - **Work**: Does it **explain** why strategies differ and outcomes differ? - **Relevance**: Does it address what participants **continually work on**? - **Modifiability**: Can you refine the core if tomorrow’s data demands it? --- ## Key references - Glaser, B. G. (1992). *Basics of grounded theory analysis*. Sociology Press. - Glaser, B. G. (1978). *Theoretical sensitivity*. Sociology Press. - Glaser, B. G. (1998). *Doing grounded theory*. Sociology Press. --- ## Companion skills - `open-coding`, `constant-comparison`, `memo-writing` - `theoretical-coding`, `theoretical-sampling`, `theoretical-saturation` - `substantive-theory` (write-up structure)