The Bactra Review   A Solution to the Ecological Inference Problem
In ecology proper, we often want to know things like, say, the odds of a jackrabbit being eaten by a coyote in a given years. But because of this, that tends to be the level at which ecological data are aggregated, rather than (in this case, just to stick to a 2x2 table): number of animals of all sorts eaten by coyotes, number of animals not eaten by coyotes, number of jackrabbits, number of animals which are not jackrabbits.