Examines case studies as research strategy and analysis method, covering instrumental, intrinsic, and collective designs. Develops within-case and cross-case analytic strategies, addresses case selection and comparison logic, and explores how case studies enable causal process tracing and generalization.
Design a multi-case study with explicit comparison logic, conduct within-case analysis with causal process tracing, write cross-case comparative findings.
A case study is not simply a detailed description of one instance — it is a research strategy built around the logic of a bounded system. You already know from advanced research design that the choice of design should be driven by your research question. Case studies are best suited to questions that ask "how" or "why" something happened in a specific context, and where the boundary between phenomenon and context is not cleanly separable. A country, an organization, a policy episode, or a social movement can each constitute a case, but only when you have defined what makes it a coherent unit of analysis with meaningful boundaries.
The most important distinction in case study design is between intrinsic, instrumental, and collective designs. An intrinsic case study examines a case because the case itself is inherently interesting — you want to understand this specific school, this specific conflict. An instrumental case study uses the case as a vehicle to illuminate a broader theoretical question — the particular case is selected because it will shed light on something beyond itself. Collective designs extend this by examining multiple cases together, comparing them to build or test theory. Most scholarly case studies are instrumental: the case is a means, not an end.
Within-case analysis and cross-case analysis are the two analytical moves available to you. Within a single case, you use causal process tracing — following the chain of events, mechanisms, and conditions that connect a cause to an outcome. This is analogous to watching dominoes fall: you are not just correlating the first tile and the last, but documenting each intermediate step and asking whether an alternative explanation could account for the observed sequence. Process tracing can confirm or disconfirm a causal account in ways that correlational analysis cannot. When you have multiple cases, cross-case analysis compares patterns — looking for what cases that share an outcome also share in their causal conditions, and what cases with different outcomes differ in.
Case selection is where comparative logic becomes most powerful and most fraught. Selecting on the dependent variable — choosing only cases where the outcome occurred — is a design error that prevents you from assessing what conditions lead to different outcomes. Strategic designs include most-similar cases (holding many conditions constant while varying the key cause), most-different cases (varying many background conditions while sharing the key cause and outcome), and deviant cases (cases that don't fit a general pattern, which are analytically valuable precisely because they are anomalous). Each selection strategy is appropriate for different inferential goals.
The misconception that case studies cannot test theories or support generalization confuses statistical generalization with analytical generalization. You do not generalize from a case to a population the way a survey would; you generalize to a theory. If your theory predicts that a specific mechanism will operate under certain conditions, a single well-chosen case that either confirms or disconfirms the predicted mechanism is evidence. Robert Yin's framing is useful here: think of case studies as analogous to experiments — a single experiment does not prove a law, but it can decisively test a theoretical proposition when designed correctly. The strength of case study inference comes from the depth and rigor of within-case analysis, not from a large N.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.