11-17-2017, 10:26 AM
(11-16-2017, 10:22 PM)Pheidippides Wrote: I will have to disagree. Sweden is measuring all of the deaths on all of its roads so it makes sense to use the total national population as the denominator.
You still want apples and oranges though. Not all of their national population uses the roads. So if you include the people that don't use the roads in case 1, it makes sense to me to do the same thing in case 2.
(11-16-2017, 10:22 PM)Pheidippides Wrote: The region is only measuring collisions that resulted in one or more death on regional roads. The actual unit of measure is different (fatalities vs. fatal collisions).
I agree this is wrong.
(11-16-2017, 10:22 PM)Pheidippides Wrote: Also, regional roads make up 20% of the road network within the region (the rest being city, province, or private) and 60% of the total length of those regional roads is in rural areas (defined arbitrarily as being in the four townships). On top of that 11% of the region's population is rural. So they are looking at a fraction of the road total road network, and the largest portion of the regional network is used by the smallest portion of the population.
Right, but changing the population denominator doesn't address this fundamental problem. It just makes it bad in a different way (and we have no real way of knowing if the 'badness' is better or worse). The way to address it is to use a metric that actually incorporates the amount of usage.