My attention was drawn today to a website called "Helmets On Heads". This website purports, under the domain helmetsonheads.org, to be a promotional campaign to raise safety and tries to look independent but industry supported. It's actually wholly owned by the industry. This should be a warnign flag to treat claims with caution - there's a monetary incentive to overstate the case, so a little extra scrutiny is justified.
Their FACTS page makes a rather odd set of claims, which you'll probably spot if you're schooled in spotting hokey numbers. Now, I do like a good hokey number, so I thought I'd feature it here. If you can't spot it, I'll explain below. I'm not the first person to spot it, but I'm an enthusiast of arithmetical weaselry.
Here's a screenshot of the facts page as it was when I looked on the evening on 9th October 2012
OK. So, at a glance, you'd think that "whoah, I should totally rush out and buy a bike helmet. But wait! There's something wrong with the numbers.
Let's tease out the two claims that I'm concerned with.
1. Roughly one in ten cyclists killed were NOT wearing helmets
2. Only 25% of cyclists wear helmets
Have you spotted it yet?
Yep, that's right. Using these figures, the rate of death for helmet wearers is MUCH higher than you'd expect. In fact, while nine out of ten cyclists killed WERE wearing helmets, 75% of people don't wear them. So the rate of death for helmet users is absurdly high.
In fact, if helmets were only an indifferent factor, you'd expect the casualties to break down 75% non-wearers, 25% wearers. What's quoted by the site is 90% wearers, 10% non-wearers - far more deaths for helmet wearers than the proportion of use would suggest.
If we actually crunch this set of numbers with ninja hokey numberism, helmet wearers are nine times more likely to be killed, yet only constitute 25% of the population. three times more people don't wear helmets, making it - very roughly - 27 times more likely that you'll die if you don a helmet. Off the top of my head. Yep, I know my hokey numbers.
Look, I don't actually think the numbers are true (though references ARE offered). They're probably nothing more than a collossal fuckup by the copywriters behind the site, but here is an object lesson in how to communicate risk terribly while still retaining a superficial veneer of credibility.
It's hard enough examining the numbers around helmet use as it is, without innumerate dingbats making an ass-backwards case right in the middle of it all.
Point and laugh everyone, point and laugh.
posted @ Tuesday, October 9, 2012 8:45 PM