Interestingly, most accidents occur below the speed limit and increasing speed limits decrease accidents, because limits are artificially low and thus almost no one heeds them, except for a few people who do who cause accidents because accidents happen when the variability of speed is greatest.

http://sense.bc.ca/disc/disc-09.htm

What speeding does generally is increase the severity of accidents and alarmingly fast (Thanks, Newton!).

Energy levels follow square law and the human body has a rather low ceiling for the amount of energy it can absorb. It makes perfect sense that you're going to see large "averages" especially when the people going most excessively over the limit are the youngest and most inexperienced.

Now consider that young people also want to live in dense urban environments where there are a lot of other cars travelling at variable speeds, what you have is a recipe for high statistical impact accidents with energy levels following a square law with a low ceiling for death.

Lets look at a rather typical case: Young family, two kids, ages (28,28,2,3) they are travelling 10 over rushing to daycare and jobs, they run into an 85 year old grandpa driving 20 under. Everyone dies, so you have 10 over associated with (75-28+75-28+75-2+75-2) or 240 years of life expectancy, now on the other hand you have 20 under associated with -10 years of life expectancy. Lets not even get into who caused the accident according to anyone but a prosecutor.

Do this study on people over 75 and you'll find that old people should drastically increase the speed of their driving. Do this study on 0-10 year olds and you'll find that we shouldn't even be driving at all. The only thing this study says is that young people speed and when they die they have a disproportionate affect on average life span. The rest of it is just asinine statistical interpretations on data that have nothing to with speeding, and everything to do with the fact that the faster you are travelling the more likely you are to die when you crash. If they wanted to make this scientifically valid they should publish a chart showing at what age you should be speeding by how much, because once you're past 2/3rd life expectancy it's advantageous to speed according to this study, but this isn't about science, it's about speed = bad, so they can tax (err.. fine) you for doing it.

It's absolutely idiotic to add those numbers given the age and driving preferences and think that speeding is counter productive. This study is as stupid as a study that says that if Bill Gates moves onto your block that everyone on your block becomes a billionaire. Yes, it works on "average" but everyone knows that you still have to go to work (you may even have to speed to get there on time) in the morning if bill gates moves to your block.

I'd much rather get in an accident going 10 over a 50 zone than 5 under in a 100 zone (especially if it's on coming traffic).

Also, in Montana after a Judge invalidated their speeding laws as unconstitutional the number of accidents dropped.

http://www.hwysafety.com/hwy_montana.htm

I'll take real world data over a computer model any day of the week.