Tuesday, October 27, 2009

Lies, damn lies and statistics

posted Thu, 28 Oct 2004

Oh, those wacky mathematicians at Muscle and Fitness Hers. Here’s what they have to say in the November/December issue:

“8%: the amount of increased activity that workers experienced when they wore business casual compared to more buttoned-up attire.”

Why, the way this is written, you would almost think that wearing business casual clothing to work means that you will be more active at work, i.e., that the wearing of business casual clothing leads to increased activity.


My favorite line from the Simpsons: “Sick on a Saturday! What are the odds? Like one in a thousand?”
Source: http://www.posterservice.com/bigpics/3569.jpg

It certainly couldn’t be that one would deliberately wear less formal clothing when one knows one is more likely to be more active, could it? It couldn’t be that, say, the day you are going to be installing a new computer and crawling around on the floor would not be the day you would wear heels, hose and a suit?

Are reporters that stupid?

Yes.

Some of them, anyhow. (Not you, David, 28 ACT.)

When I lived in Austin, the American Statesman ran a story about HIV and blood samples at the University of Texas student health center. X% of the blood samples were HIV positive, therefore, the reporter concluded, x% of the students were HIV positive as well.

I know all of you see the flaw in the reasoning.

Students who go to the student health center are not representative of the entire student population. They are -- sick. And students who have blood samples taken are really sick.

Don’t ever believe anything you read in the paper about any sort of statistics or accounting. Or any sort of numbers, actually. Reporters have no numbers ability at all and apparently no common sense.

No comments:

Post a Comment