Tuesday, July 31, 2007

The odds are you'll get it wrong

Caution: Math ahead.

OK, now that we've scared off all the journalists (the ones who really need this), go read Mark Liberman's two excellent posts on Language Log about the math of science reporting, especially in the health sciences and genetics.

His most recent, on the problem of mistaking odds ratios for relative risk, may be a bit of a thicket for some but is a must-read. And his earlier post on the numbers that should be in such stories (Why aren't they? More mathphobia from journalists or a fear readers' eyes will glaze over?) is very good, too.

While we're at it, it would be a good thing for everyone to review the languge used in relative risk. Something that is "as likely" is a simple ratio. Something "more likely" refers to the additive amount (think percentage difference). (Same idea: 3 is three times as large as 1 but two times larger.)

Don't confuse the two.

Labels: , , , ,


Post a Comment

<< Home