Decius wrote:
] Thank you for a cool headed post. I agree with almost
] everything you are saying here. There isn't much worth
] commenting on other then that safety data ought to be
] just as easily available as fuel effeciency data.
] That's something I'd support.

There is a scientific-method kind of problem with such extensive reuse of "safety data" based on "real world" incidents rather than lab testing, particularly for specific vehicles or classes of vehicles.

The problem is that when using such data, you cannot decouple the driver and the vehicle. As a result, you cannot use this data to make general statements about the safety of the vehicle itself (such as a legislator might do in an attempt to outlaw sales/use of the vehicle).

If the population of SUV owners (or even specific vehicle models) is biased relative to the general population (such as by gender, race, age group, personality, temperament, driving skill, tendency to use cell phones while driving, etc.), then any field data you collect about it is also biased.

Consider the simultaneous introduction of a new SUV model, the Galaxy, in two parallel universes, Alien and Blob. In Alien, the marketing focus group determines that today's teenagers will flock to the Galaxy in droves as a great first car. In Blob, the ad wizards select rich, retired, politically conservative men as the Galaxy's primary target market. In both universes, the unstoppable engines of marketing quickly construct a loyal community of drivers around the Galaxy. The retirees of Blob mostly haul around golf bags and bicycles, rarely exceeding 25 mph as they amble through the wide lanes of the country club, enjoying the leather accoutrements of the Galaxy. However, the poorly trained, easily distracted, risk-taking Alien teenagers quickly earn the Galaxy a bad reputation among lawmakers, who cite sharply above-average accident rates as rationale for banning the vehicle.

Here we have the same vehicle, driven by two different populations, generating dramatically different "real world" safety data which is subsequently used for various purposes.

If we're talking about setting insurance rates, the data may have some value, as long as it can be offset by a driver's personal safety history. Consider the spotless record of the married, middle-aged Alien who enjoys going camping and kayaking on the weekends.

If we're talking about restricting the general sale of the vehicle, then we have a problem. You shouldn't be using biased data to set policy for the entire population.

RE: WSJ.com - The Scarlet SUV