The Idea

The intuition says large samples generalize and single cases do not. Often the reverse holds. A big survey gives you superficial, context-stripped correlations; a deep case study reveals mechanisms, and mechanisms travel. Ethnography in one community can uncover how reciprocity, reputation, and sanctioning sustain trust, and those mechanisms apply far beyond that community even though the local practices do not.

Close study also shows which features of context matter, which is exactly what you need to say where else a finding applies. You cannot get scope conditions from data that stripped the context out.

This is not a license for lazy small-N work. You still need a theoretical framework to interpret the case, systematic within-case analysis rather than cherry-picking, and an explicit argument about which parts are mechanism and which are local particulars.

Why I Keep It

It justifies my default study design: track fewer speakers intensively across contexts rather than survey many once; fit models deeply to one context to understand the mechanism, then test breadth. It runs opposite to the inventor’s paradox (go general to solve specifics) because they target different phases: that one is for problem-solving, this one is for inference.

Tensions

• It is easy to mistake local particulars for universal mechanisms; the generalization still has to be argued, not assumed.
• Depth and breadth trade off inside a fixed budget, and the principle does not say where the line is.

Key Sources

• Flyvbjerg, B. (2006). “Five Misunderstandings About Case-Study Research”
• Ragin, C. (1987). The Comparative Method
• Gerring, J. (2004). “What Is a Case Study and What Is It Good For?”