At PB's urging, we listened to Freakanomics by Steven D. Leavitt and Stephen J. Dubner on tape across the course of several car trips. It took a while to get through, but I liked the experience.
The book was interesting - I think the authors are very good at reframing questions in new ways and looking for data sources to answer those questions. As an exercise in creativity in research I think it's brilliant. The link between abortion and violent crime is a particularly interesting hypothesis and I can't believe they actually came up with it.
That said, I feel that many of the authors demonstrations fall prey to the same crticisms they make of other research - they are almost all correlation / anecdote based. I guess I am too much of an experiementalist at heart. Sigh.