Traffic: like bumper cars or particle physics From: Ben, Oregon NMA Date: Mon, 15 Jan 1996 01:18:18 GMT [In response to the argument that in 1974 the Washington State hiway report reports that there was a 13% drop in the fatality *rate*] There's a subtle error in your logic. Yes, the fatality rate takes into account the number of miles driven (or numbers of drivers.) It gives us an estimate of the probability of harm to a given driver (if you double the number of drivers, you double the number of accidents). But, this does not take into account the fact that each of those drivers now faces twice as many cars on the road to hit and be hit by. Essentially, you needs to take traffic density into account TWICE to have predictive numbers. A simple example: A bumpercar ride with 100 cars. Suppose that the average car bumps into 10 other cars during the ride (it is a very large area, and the drivers are limited by distance, not congestion.) That represents 1000 bumps (or 500, each affecting 2 drivers) and opportunities for injury (whiplash). Whiplash occurs 1% of the time, so you have 10 cases per run of the ride. Now, suppose that you reduce the number of cars by half to 50. Does the number of whiplash cases also cut by half to 5? No. Each of the 50 cars now experiences only 5 bumps (THIS is the density effect) because the cars are spaced more widely apart. A total of 250 bumps occur. At 1% injury risk, you have only 2.5 injuries per ride, on average. The injury RATE dropped by half, while the total dropped to 1/4. A clever fellow might point out that the increased distance between cars in the thinned density case will allow them to pick up a bit more steam before colliding, increasing the injury rate from 1% to 1.2% (a 20% increase) As a result, the injury totals would drop from 10 to 3, not 2.5 Predicting traffic fatalities is complex. More complex than a simple model such as this can predict (and certainly more complex than "speed kills.") -- Ben Langlotz National Motorists Association, Oregon Chapter Coordinator oregonnma@aol.com http://www.msn.fullfeed.com/nma/