CBC: fail.

I long ago learned to treat any mainstream media news story which attempted to use statistics extremely gingerly, but this one's just too sad for words.

The CBC excitedly informs us that 32.6% of hospital beds in Quebec are occupied by current or former smokers. This, it is suggested, means "caring for smokers in hospital costs Quebec's health-care system $930 million a year" and "smokers are taking up beds and costing the Quebec health-care system millions a year".

What is sadly lacking from the article is any mention at all of the other piece of data without which the number cited means precisely nothing...

yes, chorus along with me - so, how many people in Quebec actually smoke, then?

Well, 23.3%, apparently. So it would appear there's something to this story, in that smokers seem to be taking a disproportionate amount of hospital space. Still, you'd have absolutely no way of knowing that from the story as presented by CBC. And in fact, that number doesn't account for the 'and former smokers' wrinkle, so it's entirely possible there's virtually no substance to this story at all.

(I'm not suggesting smoking has no health effects, here. Of course it does. That's been well demonstrated by years of medical studies, many of them written by researchers with a post-kindergarten grasp on the use of statistics. I'm just pointing out a particularly egregious example of the journalistic utterly meaningless statistic...)


EmmetCurran wrote on 2010-08-18 12:30:
I've been reading this blog for years and thought it was about time I joined in! :) Ben Goldacre does a good job at ripping apart these 'studies' over at http://www.badscience.net/ and his weekly Guardian spot.
adamw wrote on 2010-08-18 15:44:
Ben does science more broadly; there's someone who does something similar focussing mostly on statistics, I think, but can't remember who/where. I love Ben's columns, though.
OisinFeeley wrote on 2010-08-28 16:52:
John Allen Paulos (author of _Innumeracy_ amongst many other works) concentrates on shedding light[1] on such confusions. He has a column on ABCnews[2] named _Who's Counting_. The Ben Goldacre stuff is excellent too. 1. http://www.math.temple.edu/~paulos/ 2. http://abcnews.go.com/Technology/WhosCounting/