Obviously, I'm not against studying the demographics of religion, I think there's a lot to learn by doing so, but it doesn't do to treat religion as a demographic phenomenon. Although a religion may have demographic implications, and you can find out a lot about a religious body's relative health (though not necessarily its truth) by examining it demographically religious bodies are primarily defined by a set of beliefs.
I ran into this book review of Darwin's Cathedral: Evolution, Religion, and the Nature of Society which attempts to use an evolutionary model for evaluating why religions are or are not successful in growing in certain environments. The reviewer, Ronald L. Numbers, pins the tail on this donkey in American Scientist. I first ran into Dr. Numbers some years back when I came across his books on the history of creationism as a movement (he's not a creationist by any stretch but has written some pretty fair books about it).