I am using first-order stochastic dominance and I am wondering if I need to cite a source.
More generally, how should one cite well-known mathematical definitions -- Is it okay to cite a somewhat recent source (like a book that covers the topic) or should one go to the effort to track down where it was first introduced?
The objectives of the citation, imho are:
In recent works, you don't have a choice, because there won't be a detailed explanation in anywhere but the original article, which covers the second point as well.
For well established developments (old stuff), the original article might be unreadable, considering current standards. The figures improved considerably in the last 10 years.
Personally, whenever is easy, I do both, cite the original work AND a current accessible textbook on the subject. If the original work is too obscure, just cite the text. In these cases, point 1 is more important than point 2.
Provide one reference to a recently, quality publication -- a reference that can be plumbed back to the original idea if needed. The presentation of that first paper is rarely crucial for new work.
