First things first. Affinity maps and affinity diagrams (the two terms are used interchangeably, as is simply "the affinity") as physical things, are actually secondary in terms of significance to the much more important and inherently creative process of creating the maps or diagrams in the first place, i.e. affinity diagramming.
Affinity diagramming is fabulously useful for extracting regularity from raw and "noisy" data (and is a classic example of systems thinking, of course). The data might be lists (of words or phrases), post-its, pictures, images, photographs, handwritten notes, questions, numbers or stats - basically any kind of data at all, in fact. The regularity might be a concept, a collection of ideas, a theme or themes, trends, tendencies, proclivities, etc etc.
With affinity diagramming, its hard to tell what regularity you're looking for, because you're not trying to impose a named structure upon apparent disorder - as you do with card sorting.
Rather, the aim is to synthesise novel structure from something that apears to be unstructured, and then name that emergent regularity.