All rights girls and boys; it’s time for another lesson in statistics. Remember, statistics are a wonderful tool, but only if you know how to interpret them. You can’t depend on the interpretation by news writers. People who write the articles accompanying stats, for the most part, aren’t statisticians. So, you should be aware of a few fundamental things first.

As I said the other day, the best polls are the large well-funded polls. These polls besides having the best-trained folks running them, they can afford a decent size sample (≈1,000 to 1,500 respondents). There’s one more thing to be considered: if these people get it wrong, it’s noticed by everyone and the poll is not trusted the next time. Don’t believe me? Do you remember the VNS exit poll that predicted that Al Gore had won the 2000 election? The VNS no longer exists.

Another thing I mentioned was history effects. This graph shows how history effects have affected George W. Bush’s approval ratings. Approval ratings are subtle things. When Bill Clinton was impeached, people said that they didn’t trust him on a personal level. But the public never felt that Bill couldn’t do his job. Clinton’s approval ratings never dipped far below 50%.

Trend lines are the single most important part of the polling process. Any single poll is a snapshot. It can’t tell you anything more than what’s happening at that moment in time. However, a trend line will give a fuller picture. Public opinion will ebb and flow on any subject. If public support falls below 40% though, it’s hard to get back. Approval ratings hold to this rule harder then a 3-year old hangs on to your leg when they want a new toy. You don’t come back from a rating below 40%.

It can be spun, it can be disparaged, but the reality can’t be avoided. So, that’s why W was running away. Face it W, Americans don’t like you, they don’t trust you, and they don’t think that you are up to the job. And that leads us to the second photo: You can run, but you can’t hide.