« Ah...No. We Watched the WHOLE Thing | Main | I Think I'm Turning Beijing-ese, I Think I'm Turning Beijing-ese »
August 14, 2008
Speed Kills
...except when it doesn't.
Highway crashes kill more than 41,000 in 2007Traffic deaths in the United States declined last year, reaching the lowest level in more than a decade.
Some 41,059 people were killed in highway crashes, down by more than 1,000 from 2006.
The fatality rate of 1.37 deaths for every 100 million miles traveled in 2007 was the lowest on record, the National Highway Traffic Safety Administration said in its report Thursday.
"Every 100 million miles" means what, I wondered. How many miles did Americans travel in 2007 for those 41K deaths?
About 270 BILLION.
That's a lot of tires. And damn near a Qantas-type safety record.
Posted by tree hugging sister at August 14, 2008 11:32 AM
Comments
More meaningless statistics.
Just how did they measure those 100 million miles? I didn't tell anyone about that trip to the grocery store I made yesterday.
Posted by: Skyler at August 14, 2008 02:28 PM
Maybe they kept track of odometer readings on car registration renewalss? that would give a pretty damned acurate reading of how much people were travelling, wouldn't it?
Posted by: Creatrix at August 14, 2008 03:04 PM
Just how did they measure those 100 million miles?
Transportation departments track vehicle traffic on arterials and highways constantly, using sensors and direct counts. You can build up a decent statistical model of vehicle travel that way. This is even more accurate on toll roads, especially if they pay by card or RF ID.
Creatrix's suggestion applies only if states track odometers (my state doesn't), but it's not unreasonable to use as samples.
Another method: fuel sales. Add up total fuel sales, apply an average consumption rate per type of fuel, and you get an estimate of miles traveled.
Finally: odometers readings are recorded during maintenance (e.g., oil changes, tire rotations, or services). You have a time period and total mileage. The math is quite simple. Again, not comprehensive, but a good source, statistically speaking.
No one of these approaches is complete. Taken together, you can get a very reasonable estimate of driving patterns. I expect this "270 billion miles" is good enough to work with.
Still, one must approach statistics carefully. But this evaluation looks reasonable, because it's measured in units that apply across the board, instead of "driver-hours", or some other artficial or arbitrary term. Good statistics are straightforward, and time or distance based demoninators are quite straightforward. Think of this as "mean distance between failures", and you can see the application.
Sis, did you
Posted by: The_Real_JeffS at August 14, 2008 03:31 PM
Whoops! Cancel the last sentence.....I fergot to delete it. Sorry, Sis.
Posted by: The_Real_JeffS at August 14, 2008 03:35 PM
No. I didn't.
Never have.
Posted by: tree hugging sister at August 14, 2008 06:03 PM
Really? WOW!
Posted by: The_Real_JeffS at August 14, 2008 09:12 PM
I can vouch for that. I've never known THS to
Posted by: Skyler at August 14, 2008 10:58 PM
Says YOU.
Posted by: tree hugging sister at August 15, 2008 12:05 AM
"No. I didn't.
Never have."
Well, maybe it's about time you did! ;o)
Posted by: Creatrix at August 15, 2008 09:00 AM
I'm not that kind of girl.
::sniff::
Posted by: tree hugging sister at August 15, 2008 11:37 AM