I've just finished reading Freakonomics. Bill already recommends it, so I'll just add a few comments.
What I found really interesting about this book it the way it challenges conventional wisdom--convenient, comforting, widely-held ideas which are not necessarily, in fact, true. The authors use regression analysis--a tool which economists use to sort through piles of raw data, holding constant every variable except two, to see if and how those two variables are related--in search of a statistical basis for these well-known ideas. The analysis looks for correlation--the degree to which two variables change together--to find relationships in the otherwise random-seeming data. The value of one variable can be an indicator--a specific measure that helps predict a more general condition--of what the other value will tend to be (all else being equal, which is what the regression analysis is trying to control). But as is often said, correlation does not imply causation, causation being that the presence of the first value makes the second value present as well.
So the book is looking for statistical correlation to support or refute conventional wisdom, and probabilistic proof of causation, showing that one characteristic or event causes another. Not too surprisingly, the book examines certain pieces of conventional wisdom that turn out to be wrong; it shows how the data doesn't support any correlation, and finds where there is a correlation, often a surprising one, one that can often be shown to be a cause.
A lot of the book hovers around the motivations--what economists call incentives--for why we do what we do. Of particular interest is a section on why we worry about problems that are rather unlikely--airplane crashes, terrorist attacks, mad-cow disease--yet sometimes don't worry about problems that are much more likely--car crashes, heart disease, E. coli in your own kitchen. A lot of it has to do with suddenness, control, and dread. A terrorist attack happens suddenly, is beyond our control, and seems like a horrifying and needless way to die--thus our fear of terrorism, even though any one of us is very unlikely to die of a terrorist incident. By comparison, heart disease builds up over years, is caused primarily by the diet we choose to consume, and is seen as more of a natural consequence of aging--thus we don't fear heart disease so much even though it's a very common and preventable cause of death.
One consultant on risk communication even has a formula for what motivates us: Risk = hazard + outrage. Because outrage tends to have a greater weight, unlikely hazards can seem very risky if they're shocking enough. Activists try to make an issue seem more outrageous; keepers of the status quo try to downplay outrageousness.
The last two chapters concern what makes a good parent. Surprisingly, the data shows that the eventual success of your child has a lot more to do with who you are as a parent--education, steady job, optimism, etc.--then what you do with your child--read to them, spank them, take them to museums. How well your child will turn out, at least early in their life, mostly has to do with how well you turned out before your child was ever born. So if you made a mistake with your child today, cut yourself some slack, it probably hasn't screwed them up that much after all.