by I’ve been doing research on my next car purchase, and have been specifically focused on safety recently. Much to my surprise, the safety ratings auto manufacturers brag about and use to market their cars often have little correlation with real world data. Tests, like polls in elections, are useful tools, but what actually happens to car drivers and occupants is as a real and as import as outcomes on election day.
Don’t get me wrong, models and tests are not without use. You can’t just design and build a car and wait to see what happens people drive it. But there are limitations. Tests give an approximation of outcomes based upon various scenarios within the a range of given constraints. But eventually you need to take a look at the eventual outcomes. No campaign manager gets praised for winning polls. Election day is the real deal. But for some odd reason, it doesn’t really work like this for cars. The five star safety rating that sub-compact you are driving is awarded on the basis of a lab test, and in all probability, it really has nothing to do with actual road safety as reported by insurance companies.
Three days ago I really had no reason to doubt what I was told about car safety. I’m not a big car guy. I’m not an automotive engineer. I really never put much thought into it until recently, when I started doing some deep dives into the data, but car safety ratings seem to be yet another case professional quackery we seem to be surrounded by in our culture. At times, every which way we turn, information and feedback about our world is debauched.
Just tell me how many people get killed per some large sample of miles driven. And, although that too is an imperfect and incomplete measure, it must be better than a laboratory. For example, a car a family member drives is #3 on the list of unsafe cares, in terms of fatalities per 100 million miles driven. To find a car slightly more unsafe, one would have to drive a Corvette, which isn’t exactly designed for a family of four to travel to the lake or church on weekends. In another case, a car with a three (of five) star safety rating reported zero fatalities last year, despite a large volume of those cars being sold and driven. You would think having zero fatalities associated with the car would count for … something. By the time I got to researching the fifth or sixth car, and compared industry safety ratings to reported and actual real world safety outcomes, it quickly became abundantly clear that the industry safety ratings had very little connection to reality. I’m not sure what is shocking — that fact, or that I’m shocked.
And, of course, there’s more. A few conversations, with a few different people about this topic, revealed something I’ve observed in other situations. People really want to believe the model the esteemed experts sell them. They have an extremely hard time making sense of actual data, and determining what it means on their own. And if the conclusions they draw differ from the experts, even when they are confronted with numbers that are literally of life and death in nature, they defer to authority. They want to believe there are “white hats” out there, looking out for them. And if some information comes their way, which is at odds with that belief, they simply ignore it. And they are eager to do so. Because what’s the alternative? Its just too much to ponder, I suppose.
