Lying with statistics
I just hate that phrase. HATE. It makes it seem like the use of statistics is particularly suited to bamboozling people* -- this despite the fact that most people don't remember middle or high school math after they graduate (or don't, as the case may be). In a bastard's hands, sure, statistics can deceive (though it's hard to actually "lie" with them). But then, in a whistle-blower's hands, they can unmask unjustices, such as how an executive's golden parachute raised the average earnings (or whatever the proper finance term is) at his company for that year. We thus return to a boring truism: statistics don't deceive people -- people deceive people. We root out deception by cracking down on human mendacity, not by regarding "mere statistics" as the charlatan's art, as in the popular, smug dismissal that "you can prove anything with statistics."
In the end, though, most deception in the world doesn't involve the use of numbers -- again, if the audience can't comprehend the myth, they can't believe in it, and most people want to involve numbers in their daily lives like they want to sit in the dentist's chair every morning and evening. It will always be easier to bullshit using words rather than numbers, since our species' evolved method of communication is linguistic -- basic numeracy has not been widespread for more than probably 100 years (if that), so there is no evolved knack for lying with numbers, whereas there is a wealth of tricks wired into the inveterate huckster. Therefore, intellectual self-defense really should focus more on taking apart verbal arguments.
That said, you'll still come across bad statistical arguments more frequently than you'd like to, and so Mark Chu-Carroll from the ScienceBlog Good Math, Bad Math has started a nice "back to basics" series on statistics. So far, there are entries on mean, median, and mode and the normal distribution.
* A joke among topologists is their self-description as those who can't tell a coffee cup apart from a donut, and yet based on their wacky wisdom, we hardly conclude that "there are lies, damned lies, and topology."