theoretical-saturation

# Theoretical Saturation **Theoretical saturation** is the stopping rule for **category development** in grounded theory: you keep sampling and comparing until additional data **do not** change your understanding of a category’s **properties**, **dimensions**, or its **relationships** to other categories **in ways that matter** to the emerging theory. Saturation is **not** “I interviewed enough people.” It is **theoretical**. Use this skill when planning sample closure, responding to deadlines, or evaluating whether a category needs **more** comparative work. --- ## What saturation means (classic GT) **Saturated category**: further comparison yields **no new properties** and **no new hypotheses** about relationships relevant to the **core story**—only more examples of the same variation pattern. **Important nuance**: you may always collect **more stories**. Saturation asks whether those stories **refine** the conceptual structure. --- ## What saturation does not mean - **Verbatim repetition** of participant language. - **Exhaustion** of all possible human experiences in a domain. - **Statistical** representativeness. - **Comfort** or **fatigue** of the researcher. - A fixed **minimum n** (sample size cannot substitute for comparative logic). --- ## How to assess saturation (procedure) ### Step 1 — Specify the unit of saturation Saturation applies to **categories** (and key relationships), not to “the dataset” as a blob. Ask: “Which **category** am I evaluating?” (e.g., *soft framing*). ### Step 2 — Track properties and dimensions For the focal category, list: - **Properties** (what kind of thing it is) - **Dimensions** (along what axes it varies) Update this list as comparison proceeds. ### Step 3 — Run targeted comparisons Bring in new incidents **explicitly chosen** to test weak spots (theoretical sampling). ### Step 4 — Look for stabilization signals **Saturation signals**: - New incidents **instantiate** known properties without refinement. - Negative cases **fit** refined boundary statements rather than exploding them. - Memo outline **stops shifting** for that category’s role in the core story. **Non-saturation signals**: - You still cannot state **when** the category appears vs not (boundary thin). - You have **unexplained** deviance stockpiled “for later.” - Each new interview changes **definitions** materially. --- ## Premature vs genuine saturation ### Premature saturation (common) **Causes**: - Homogeneous sample (same role, same site, same ideology). - Shallow interviews (no incident detail). - Avoidance of **negative cases**. - Confusing **story repetition** with **conceptual redundancy**. **Symptoms**: - Thin boundaries (“always/never” claims). - Codes that **collapse** distinct processes. - Surprises late in fieldwork that **rewrite** the core. **Corrective actions**: - Theoretically sample for **deviance** and **contrasting contexts**. - Use **incident-based** probes to deepen properties. ### Genuine saturation (ideal) **Markers**: - Variation is **mapped** as dimensions, not as chaos. - Deviance is **accounted for** by conditions/strategies. - Additional data **feel** redundant **conceptually** (even if narratively fresh). --- ## Indicators at different levels | Level | What to check | Saturation hint | |------|----------------|-----------------| | Code | Definition stable; splits resolved | Merges hold across new data | | Category | Properties/dimensions stable | New incidents align to matrix | | Relationship | Hypothesis stable under tests | Contradictions explained by boundaries | | Core story | Integrated outline stable | Selective coding yields marginal gains | --- ## Dey’s “theoretical sufficiency” (useful complement) Some authors (including Dey) argue “saturation” language can overclaim. **Theoretical sufficiency** reframes the question: > “Have we developed categories **adequately** to support a defensible argument for this study’s aims?” **Practical use**: - In bounded dissertations, sufficiency acknowledges **scope limits** while preserving rigor. - Pair with transparent **negative case** accounting and **audit trails**. This does **not** mean lowering standards; it means **honest** alignment between claims and evidence. --- ## Checklist for assessing saturation ```text Category / relationship under review: Properties listed (complete?): Dimensions listed (complete?): Boundary statement written in one paragraph: Negative cases on hand (count + sources): Unexplained deviance remaining (list): Last time this category changed meaning (date + why): Planned next comparisons (if any): Decision: continue sampling / close category / reopen category ``` --- ## Writing saturation claims in manuscripts A strong saturation claim typically includes: - **What** categories/relationships saturated - **How** sampling targeted variation and deviance - **What** would have changed the theory if new properties appeared Avoid vague “saturation was reached” without **category-level** specificity. --- ## Relationship to theoretical sampling Sampling and saturation are **paired**: - Sampling finds the **next** comparative test. - Saturation tells you when **additional tests** stop paying conceptual rent. --- ## Key references - Glaser, B. G. (1978). *Theoretical sensitivity*. Sociology Press. - Glaser, B. G. (1992). *Basics of grounded theory analysis*. Sociology Press. - Glaser, B. G. (1998). *Doing grounded theory*. Sociology Press. - Dey, I. (1999). *Grounding grounded theory*. Academic Press. (discussion of sufficiency vs saturation framing) --- ## Companion skills - `theoretical-sampling`, `constant-comparison`, `memo-writing` - `selective-coding`, `substantive-theory` (fit, work, relevance, modifiability) - `glaserian-grounded-theory`