4 ms·

You seem to imply that accident rates will be proportional to miles driven. I doubt that is true. I would expect the graph of the "chance of an accident per m

You seem to imply that accident rates will be proportional to miles driven. I doubt that is true. I would expect the graph of the "chance of an accident per mile" vs "miles driven" to look something like this: 0 | * | * | * | * | * | * | * * * | - - - - - - - - - - - - - - - minimum accident rate | ---------------------------------------------- 0 I have two reasons: First, as miles driven increases, driver experience increases. Accidents are much more common among (I'm guessing) 16 year olds with a freshly minted license than a 40 year old. Second, as miles driven per year increases, the percentage of driving time spent on the freeway increases (due to taking long trips, presumably the major source of milage increase), but accidents are less likely on the freeway. This argument is admittedly much more specious than the first. Following this reasoning, it is not at all clear that "the improvements in safety are even more dramatic if you consider the average person is driving about 3000 more miles per year than they were in 1975" because those extra miles may not be the most dangerous miles.

Zak 15y ago
First, as miles driven increases, driver experience increases. I suspect this effect may not be as significant as you think. Certainly, someone who has been driving every day for 20 years will usually be a better driver than someone who has been driving every day for 20 days. I suspect the difference between 2 years and 20 isn't all that dramatic though. I further suspect that most driving is not practice, by which I mean that drivers are not actively thinking about their driving and how to be better at it. Most of us, once we're reasonably proficient stop actively thinking about the act of driving entirely unless an unusual situation occurs. If your second reason turns out to be correct, then it also means that a larger percentage of driving than before isn't useful practice for avoiding crashes since highway driving tends to be safe and predictable.
nkurz 15y ago
You make good points, but I think that "average miles driven" is likely just the total miles driven divided by population. I'd guess that many (most?) of these new miles are because a greater percentage of the population has a car than in 1975. There may even be relatively more <3000 mile per year drivers now than before. I don't think the linked data is quite enough to strongly support either interpretation. Still, the drop looks more impressive on a per mile basis than on a per person basis.

Zak 15y ago

First, as miles driven increases, driver experience increases. I suspect this effect may not be as significant as you think. Certainly, someone who has been driving every day for 20 years will usually be a better driver than someone who has been driving every day for 20 days. I suspect the difference between 2 years and 20 isn't all that dramatic though. I further suspect that most driving is not practice, by which I mean that drivers are not actively thinking about their driving and how to be better at it. Most of us, once we're reasonably proficient stop actively thinking about the act of driving entirely unless an unusual situation occurs. If your second reason turns out to be correct, then it also means that a larger percentage of driving than before isn't useful practice for avoiding crashes since highway driving tends to be safe and predictable.

nkurz 15y ago

You make good points, but I think that "average miles driven" is likely just the total miles driven divided by population. I'd guess that many (most?) of these new miles are because a greater percentage of the population has a car than in 1975. There may even be relatively more <3000 mile per year drivers now than before. I don't think the linked data is quite enough to strongly support either interpretation. Still, the drop looks more impressive on a per mile basis than on a per person basis.