Case study analysis focuses on a small number of cases expected to provide insight into a causal relationship across a larger population of cases. A case selection based on representativeness may not generate revealing insights. Researchers, therefore, prefer information-oriented sampling, as opposed to random sampling, for selecting case-study subjects.

Cases can be selected based on three criteria:
1. Key cases: The case might be given and studied with an intrinsic interest in the case itself or the circumstances surrounding it. The researcher has no interest in generalising the findings. He focuses on understanding the case. If findings are generalised, it is done by audiences through “naturalistic generalisation”.
2. Purposefully or analytically selected case: A case may be purposefully selected in virtue of being, for instance, information-rich, critical, revelatory, unique, or extreme (Stake 1995, Patton 1990). In this case, then there may be an interest in generalising the findings.
3. Local knowledge cases: Cases are chosen because of researchers' in-depth local knowledge. This means researchers are in a position to “soak and poke” (Fenno, 1986), and thereby to offer reasoned lines of explanation based on this rich knowledge of setting and circumstances.

* * *

Sources:
http://www.kenbenoit.net/courses/principles/Gerring_Case_Study_Research_Ch5.pdf
http://www.psyking.net/htmlobj-3839/case_study_methodology-_rolf_johansson_ver_2.pdf
http://en.wikipedia.org/wiki/Case_study