Uri Bram, a recent Princeton graduate, has just published an ebook called Thinking Statistically. The book is aimed at conveying a few important statistical concepts (selection bias, endogeneity and correlation vs. causation, Bayes theorem and base rate neglect) to a general audience. The official product description:

This book will show you how to think like a statistician, without worrying about formal statistical techniques. Along the way we'll see why supposed Casanovas might actually be examples of the Base Rate Fallacy; how to use Bayes' Theorem to assess whether your partner is cheating on you; and why you should never use Mark Zuckerberg as an example for anything. See the world in a whole new light, and make better decisions and judgements without ever going near a t-test. Think. Think Statistically.

Less Wrong members will be familiar with these topics, but we should keep this book in mind as a convenient method of getting friends, relatives, acquaintances, and others interested in understanding rationality.

Eliezer's An Intuitive Explanation of Bayes' Theorem gets a shout-out in the Recommended Reading at the end.

$3? Purchased. I'll skip a Rockstar somewhere.

Also recommended (as a refresher/introduction) - Think Stats, which offers a free download. The book is specifically oriented towards programmers, and many of the exercises involve writing little python programs to operate on some data sets.

As far as I can still (not a statistician!) it has all the important basics - probability distributions, hypothesis testing, bayes' rule, but also graphing your results.