open-coding

# Open Coding (Classic Grounded Theory) Open coding is the **first major analytic move** in classic grounded theory: you fracture qualitative data into **incidents** and label them with **codes** that stand for **conceptual** meanings. The goal is not to summarize paragraphs but to **generate concepts** that can be compared, refined, and later integrated around an emergent **core category**. Use this skill whenever you are beginning analysis, returning to fresh data after a break, or deliberately **re-opening** coding after discovering a better fit. --- ## Purpose and outcomes **Purpose**: Turn raw data into **discrete conceptual indicators** that can enter **constant comparison**. **Typical outputs**: - A working **code list** (codes with short definitions and examples). - **Annotated data** (tags, margin notes, or software-linked segments). - **Memos** capturing hypotheses about meaning, relationships, and puzzles. - **Early categories** (higher-level groupings of codes with properties/dimensions beginning to appear). --- ## Line-by-line and incident-to-incident technique ### Line-by-line (dense passages) For rich narrative segments, move **line by line** (or sentence by sentence) asking: 1. **What is going on here?** (substantive meaning) 2. **What category(ies) does this indicate?** (conceptual label) 3. **What other incidents compare/contrast?** (comparison cue) Do **not** code “the whole paragraph” as one blob unless it truly expresses **one** incident. ### Incident-to-incident comparison An **incident** is a **meaningful chunk** of data that can be compared: an event, action, feeling, statement, turning point, or interactional move. After coding an incident: - Ask: “What is this incident like? Unlike?” - Pull **another incident** (same interview, different interview, different source) and compare at the **conceptual** level. - Refine code definitions when comparisons reveal **properties** (attributes) and **dimensions** (ranges). --- ## Types of codes you will use ### Substantive codes Conceptual labels derived from **what the data is about**. Examples (illustrative only): *shielding credibility*, *deferring decisions*, *patching workflow*. These should be **abstract enough** to travel across interviews, yet **faithful** to data. ### In vivo codes Use participants’ **own terms** when they **condense meaning** powerfully. In vivo codes keep you close to **emic** language while still enabling comparison. **Caution**: Not every catchy phrase is conceptual. If an in vivo term is **too idiosyncratic**, translate it into a **more general** substantive code while preserving the participant language in a memo. ### Gerund coding (“-ing”) Glaser often recommends **gerunds** to keep **process** in view: *managing time*, *containing conflict*, *signaling competence*. This helps you see **action/interaction** rather than static nouns. ### Provisional “hypothesis codes” Early codes that may be wrong are fine. Classic GT **expects** code churn. Rename, split, merge—**document** changes in memos. --- ## Fracturing data (practical moves) - **Break on shifts**: new topic, new actor, new emotion, new tactic, new consequence. - **Break on strong lines**: especially vivid, repeated, or consequential moments. - **Break on comparisons** you can already imagine: “This sounds like X, but also unlike X because…” - **Tag “don’t know yet”**: if unsure, use a temporary label + memo the uncertainty. --- ## Worked example (abbreviated) **Data excerpt (fictional)**: > “I didn’t tell my manager about the side project at first. I needed to see if it would even work. Once it looked real, I scheduled a short chat and framed it as ‘experiment’ not ‘commitment.’” **Possible open codes (illustrative)**: - *delaying disclosure* (substantive) - *probing viability privately* (substantive) - *softening risk framing* (substantive; gerund-friendly: *framing-as-experiment*) - *staging legitimacy* (hypothesis code—might merge/split later) **Comparison prompts**: - Compare to another incident where the participant **disclosed immediately**—what conditions differ? - Compare to an incident of **never disclosing**—what consequences differ? **Memo seed**: > “Hypothesis: disclosure timing is managed through **viability thresholds** + **linguistic risk reduction**. Need more negative cases where disclosure backfired.” --- ## Output format (recommended) ### 1) Code entry template ```text Code name: Definition (1–3 sentences): Inclusion criteria: Exclusion criteria (common confusions): Example incidents (2–3 brief quotes with source IDs): Related codes (merge/split notes): Open questions: ``` ### 2) Session log (audit trail) ```text Date / dataset: Incidents coded (count) + IDs: New codes added: Codes renamed/merged/split (why): Memos written (titles): Next comparison targets: ``` --- ## Quality checks during open coding - **Abstract enough** to compare across cases, **concrete enough** to trace to incidents. - **Avoid laundry lists** of topics; aim for **analytic** codes. - **Memo** whenever you claim a relationship (“because,” “in order to,” “leads to”). - **Seek negative cases** early: incidents that seem to contradict your “favorite” code. --- ## When to move toward selective coding You are **not** finished with open coding in a single week; it matures through comparison. Move toward **selective coding** when: 1. A **core category** candidate **recurs**, **explains** a wide swath of variation, and **connects** many other categories. 2. You can **delimit**: additional open coding feels **redundant** relative to the emerging integrated story. 3. **Theoretical sampling** questions become **targeted** at the core and its related conditions/strategies/consequences. If you move too early, you risk **forcing** a core. If you move too late, you risk **endless sprawl**. Let **comparison + memo sorting** guide the transition. --- ## Common pitfalls - **Paraphrasing** instead of conceptualizing. - **Over-coding** with synonyms; **merge** aggressively after comparison. - **Under-memoing**; losing the trail of theoretical reasoning. - **Theme sorting** without comparison (categories that “sound nice” but don’t explain). - **Premature literature import** that names your phenomenon before you’ve earned concepts. --- ## 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 - `constant-comparison`, `memo-writing`, `theoretical-sensitivity` - `selective-coding` (next major transition) - `glaserian-grounded-theory` (full process map)