Remember kids... always wear a helmet. (The almighty bikeradar helmet thread)

I make no report to the "goodness" (academic term) of that article (the journal one; the Star one is garbage). However, in my experience, regarding sample size, you don't get published without it being large enough to make statistically relevant claims.

In regard to ignoring other variables - no idea.

(Did MSc level quants 5 years ago. I forget all of it. And hated it then, anyway).

Sorry didn't mean to seem like I was arguing with you, merely pointing out the flaws as I see them (from my biased position of not wanting a helmet law and not wearin them myself except when forced to), he claims 95% statistical accuracey, which is the standard. But that is based on the assumption that the reports contain all the data required. If the Canadian reports are anything like British ones I doubt they do.

I've not done masters level stats for 15 years! Although I do find it interesting...

Anyway, by comparison, the most often quoted data about injuries and helmets had a sample size of around 3,500 if I remember correctly... As oppossed to this one of 127. Make of that what you will.

The cynical side of me reckons if you peel the funding onion back far enough you might find bike trade money somewhere in there...

One thing that I've always suggested as proof that helmets aren't up to much is the fact that helmet manufacturers are not funding research.

From the style of the question and the writing in the paper I'd suggest he believes helmets saves lives and wants to prove it, but that's my biased opinion.