Recommended by a colleague at work, this appears to be one of those insta-classic works. From it, I have learnt that economics isn't about monetary policy, and things associated with that, but is, in fact, about performing statistical analysis on anything you like to get correlations, and then identifying incentives to work out causation. But I'm being a little facetious.
Despite the name, it's not economics in the traditional sense. The book's about challenging conventional wisdom through statistical analysis. Effectively a set of papers munged into beach reading, it's a decent set of anecdotes about what data analysis can show. It's also, if you don't just ignore the method and take the results at face value, a call-to-arms for rational, critical thinking across broad areas of policy.
The book is not perfect. It's almost amusingly full of American-hot-button-pushing issues (such as abortion), and I think this contentiousness doesn't do the book any favours, even if it does for the book's press. It's also slightly alien for non-US readers. But as a layman's demonstration of the power of statistics, it's great.