Originally Posted by
wphamilton
The problem is the data. Don't get me wrong, I've done similar calculations from a lot of different sources and I come to a similar conclusion. But calculations like this can only give you a very general idea of the magnitude of risk, with a large range and low degree of confidence. I'm naturally drawn to it because I'm a software engineer in the risk management industry (and other information) with a degree in Mathematics. Just sayin' so someone doesn't call me an illiterate moron blowing smoke - a hazard in this thread.
Sure the data accuracy and completeness is suspect, but that's by no means the only problem. It's also necessary to properly use the data one has in subsequent calculations.
In the post you cite by bragi he tries to calculate the risk of a single individual having a severe enough accident to warrant an ER visit. To do so to first order one should divide the number of ER visits nationally by the total number of people participating in the activity - but bragi doesn't do that. He divides by the number of bicycle trips (4 billion in a year) rather than by the number of people. Since most bicyclists will take many bike trips each year the conclusion is way off. Especially applying it to car-free bike enthusiasts who probably take hundreds of bike trips per year.
But it's ok since after making this error to underestimate by a couple orders of magnitude he then applies the 88% effectiveness of helmets from the widely discredited Harborview study. As meanwhile just got done showing with several references, the actual number here should be about 3% +/- 3%, so we have a second error of about 1.5 magnitudes in the other direction so they pretty much cancel each other out.