Efficiently Studying Rare Events: Case-Control Methods for Sociologists

Abstract

Case-control designs involve samples stratified disproportionately on a binary dependent variable. This design, though infrequently used by sociologists, offers tremendous logistical efficiency in the study of rare events, such as divorce, joining a religious cult, or committing a crime. This paper attempts to sensitize sociologists to the many situations in which this design is useful, and offers general and accessible guidance on the practice of case-control research. Using the epidemiologic literature, I explain the underlying logic of case-control design, discuss how to conduct case-control sampling, and briefly cover data analysis issues. I conclude with an empirical illustration of a case-control study examining factors associated with the change of chief administrative officers at post-secondary educational institutions.