> You really need to look at incident rates rather than totals when making comparisons across categories like this. Depends on what you want to figure out. For

> You really need to look at incident rates rather than totals when making comparisons across categories like this. Depends on what you want to figure out. For finding out the comparative danger of an individual doing activity A versus activity B, sure. But if the group doing something exceedingly dangerous is quite small, it might not warrant an intervention as far as public health etc is concerned. Scale matters (sometimes). On the flip side, you can of course argue that the individuals who just happen to be in proximity of the dangerous activity are being unfairly exposed to risk they haven't consented to (if the activity poses a danger to bystanders, that is). We don't let people do whatever they want just because the people it affects are few, either.