How is a five-point lead a "statistical dead heat"?

I'm glad I was not the only one who was extremely confused when I clicked on the headline claiming that CNN's latest poll showed Obama and McCain in a "statistical dead heat," only to find that Obama was still up five points. Nate Silver over at fivethirtyeight.com did some research on the topic, and it turns out that the National Council on Public Polls recommends that "it should not be called a "dead heat" unless the candidates are tied with the same percentages." Of course, calling the race a "statistical dead heat" did get us to click on the link to the story, which was CNN's goal . . . but it really does seem like a bit too much of a stretch.

Also, I'd like to take this opportunity to recommend fivethirtyeight.com wholeheartedly. The founder is a member of Baseball Prospectus, stat geniuses who have helped revolutionize the way people look at baseball (see the excellent book "Moneyball" for more on that movement). With 538 (which is named after the total number of electoral votes), Silver brings that level of stat-obsession to politics, creating a whole new set of interesting ways to look at the election. For example, he calculates things like the odds of Obama losing Ohio, but winning the election, Obama and McCain winning in landslides, and even the possibility of an electoral college tie at 269. It's becoming a must-read for the 2008 election.