Matching methods of various sorts have become a hot topic in the social sciences lately, in part because they offer a means of untangling causal effects from observational data (see, e.g., here).
Now the latest issue of Statistics in Medicine has an interesting paper by Peter Austin surveying the application of propensity score matching in the medical literature between 1996 and 2003. The paper is followed by a series of illuminating responses to the paper, and a rejoinder by Dr. Austin.
Beyond what the papers say about the (possible mis)use of such techniques by medical researchers, they collectively provide a nice conceptual overview of propensity score matching (and matching methods in general). Particularly clear is the discussion of balance tests (of the sort outlined here), which Austin argues are central to the correct use of such methods.