case-study-methods

# Case Study Research Methods A **case study** examines a **bounded system** (a case) using multiple sources of evidence. It is suited to **“how”** and **“why”** questions where the researcher has little control over events, and context matters. Robert Yin and Robert Stake offer influential, partially complementary frameworks. ## Yin’s logic of the case study Yin emphasizes **rigor through design**, not just data collection. A case is an empirical unit (e.g., organization, program, community, event) analyzed in depth. ### Single vs multiple case designs - **Single case** — Justified when the case is **critical**, **unique**, **revelatory**, or **longitudinal**. Risk: weak generalization. - **Multiple case** — **Replication logic** (literal or theoretical): cases are chosen to predict similar results (literal replication) or contrasting results for predictable reasons (theoretical replication). Strengthens analytic generalization to theory, not statistical populations. ### Holistic vs embedded - **Holistic** — The case as a single unit of analysis. - **Embedded** — Subunits (e.g., teams within a firm) analyzed; requires clarity about **how subunit findings aggregate** to the main case. ## Case selection Cases should align with **research questions** and **propositions** (if used). For multiple cases, define **selection criteria** before data collection when possible to reduce cherry-picking. Document why cases were included or excluded. ## Theoretical propositions Optional but powerful: explicit expectations guide **data collection priorities** and **analytic focus** (e.g., “Implementation success depends on leadership support”). Propositions differ from hypotheses in quant studies—they orient inquiry while allowing qualitative refinement. ## Data collection in case studies Typical sources: - **Interviews** — Key actors, triangulated roles. - **Documents** — Minutes, policies, reports, communications. - **Direct observation** — Meetings, workflows, artifacts in use. - **Archival records** — Metrics, histories, legal filings. Yin stresses **converging evidence** (triangulation) and a **case study database** (organized raw data, logs, and analysis notes) separate from the final report. ## Chain of evidence Readers should be able to trace **claims → displays → data**. Maintain: - Citation to specific sources (interview ID, document type, date). - Clear protocols linking evidence to assertions. - Analytic memos documenting interpretive leaps. ## Analytic strategies (Yin) - **Pattern matching** — Compare empirically based patterns to predicted ones; may involve **rival explanations** ruled out by evidence. - **Explanation building** — Iterative narrative linking causes and outcomes; guard against researcher bias through triangulation and disconfirming evidence. - **Time-series / logic models** — When temporal ordering or program theory is central. ## Cross-case synthesis (multiple cases) For each case, produce a **standalone description** and **within-case analysis**. Then conduct **cross-case comparison** using a uniform framework (e.g., tables of constructs). Look for **patterns**, **contrasts**, and **theoretical replication** support. ## Stake’s approach: intrinsic vs instrumental ### Intrinsic case study The case itself is of primary interest (e.g., a particular school). Findings emphasize **understanding that case**, not broad laws. ### Instrumental case study The case is examined to **illuminate a broader issue** or refine a theory. The case remains particular, but the lens is **instrumental** to external understanding. ### Collective case studies Several instrumental cases shed light on a phenomenon; resonates with Yin’s multiple-case design but retains Stake’s emphasis on **experience** and **interpretation**. ## Stake’s emphases - **Thick portrayal** of cases and **reader naturalistic generalization**. - Attention to **emic views** and **interpretive rigor** rather than only deductive propositions. - **Progressive focusing** — Early field visits inform what matters next. ## Yin vs Stake (practical contrast) | Feature | Yin | Stake | |---------|-----|--------| | Emphasis | Design, replication, propositions | Interpretation, portrayal, experience | | Generalization | Analytic generalization to theory | Naturalistic, reader-led | | Rigor | Database, chain of evidence, methods triangulation | Interpretive discipline, reflexivity | ## Reporting - Clear **case boundaries** and **context**. - **Methods** section with data sources and analysis steps. - **Within-case** chapters/sections before **cross-case** synthesis. - Explicit **limitations** and **rival interpretations** considered. Use this skill when the user designs multi-site evaluations, organizational studies, or compares Yin and Stake for dissertation methodology.