This piece from New Geography looks at the trouble with demographic models and the potential problems that can arise when some pieces of data are wrong.
"The first step in planning for the future is to get a picture of what the terrain is going to look like when you get there. That's mainly what I do for clients, audiences and subscribers, and demographics provide the frame (like assembling all the straight-edge pieces of a jigsaw puzzle first). But here's the thing about projections: a small change at an inflection point, or the inclusion (or exclusion) of salient variables, can result in big changes to the future you are trying to describe. So like all treatments of the future, everything depends on the underlying assumptions, and the salience of the variables chosen for inclusion."