Monday, September 26, 2005

A statistically insignificant protest

There were some anti-war protestors in DC this weekend, we're told. About 100,000 according to the reports I've seen. The media is breathlessly reporting it, no doubt, as a sign of the public's decreasing support for the US mission in Iraq.

Let's take a deep breath. Poll questions are usually asked as yes/no questions - it's easier to parse out for public consumption. But another method is to use a Likert scale - the familiar five-point strongly agree / agree / neutral / disagree / strongly disagree.

It's safe to say that 99% of the protesters in DC this weekend would fall on the "strongly disagree with the war" end of the scale. After all, they gave up a weekend (which they could have spent getting pre-positioned to provide relief services to hurricane victims) and no small amount of money. (DC is not a cheap place to stay, and I don't think they all camped out on the Mall or found crash space with symathetic locals.)

So these folks are committed to their cause. Customer service experts tell us that for every customer who complains, there are ten just as dissatisfied who don't take the trouble to complain. Let's give that number to the protestors and bump it by an order of magnitude. Let's assume that for every person on the Mall waving signs like "Bush is the terrorist" or "Socialist Revolution is the Only Answer" or "We support our troops when they shoot their officers" there are 100 people back home (or getting pre-positioned to help the communities devasted by Rita and Katrina) who support their cause, who are strongly opposed to our mission in Iraq.

So that rally on the Mall represents 100 x 100,000, or 10^2 x 10^5 or 10^7. 10,000,000. Out of a country of 300,000,000. Ten out of three hundred, one out of thirty.

That's 3.3% - the sampling error for most national polls. Way out on the tail of a normal distribution, beyond two standard deviations. Statistically,the anti-war protestors are indeed out on the fringe.

1 comment:

Pete-Detroit said...

Nice - radically inflated stats and they're still insignificant. Brilliant!