The term “risk set” is a common way to refer to a population of people who might, but won’t necessarily, be described by some event: Members of a particular protest group are in the set of people who might reasonably be expected to participate in a given protest; people who work at the mid-town mall are in the set of employees who might be affected by rezoning; and so on. While the term comes from survival analysis, where it often describes a population genuinely at risk for something dangerous (heart attacks, diabetes, other health outcomes), it has been adopted more broadly as a social science shorthand, and doesn’t really imply any kind of real “risk” any longer.
I learned yesterday that, when speaking to the good people in Human Subjects Protection1 about the generation of a sampling frame, it’s good to not assume that they know this distinction and just to shy away from the r-word altogether. Makes everybody much more comfortable that way.
1 They really are good people, having helped me to be very clear about the methods by which I want to collect data. They’re just very, ah, precise.