So we bought H. a baby seat, because, you know, the state troopers told us we couldn’t strap him to the hood of my truck anymore. Car seats are kind of a bummer — they’re marketed as the difference between life and death for your baby, but ONLY if they’re installed correctly. There’s a common meme that floats around the internet that “[X]% of child safety seats are installed incorrectly!” — where X is some alarmingly high number, usually between 75 and 90.
Now, if that turns out to be true, I think we can safely say that the problem lies not with installers but with the manufacturers. If 90% of seat belts were worn “incorrectly,” I don’t think we’d be satisfied with manufacturers sending us to police stations to get properly belted in.
But where does that statistic come from? Top Google results for “car seats installed incorrectly” cite the National Highway Traffic Safety Administration, but none of them link to it. (They also cite wildly different numbers.) It’s true that the NHTSA throws around the figure “3 out of 4” in at least a couple of places. But nowhere does that agency bother to cite a specific study or explain how it arrived at that number.
A few state law enforcement agencies have provided real statistics about car seat installation shortcomings: Pennsylvania inspected 619 vehicles in September of 2008 and found that 80% of those were installed incorrectly; Nebraska, inspecting 7,900 vehicles over the course of a year, found an almost unbelievable 89% failure rate.
But are these numbers representative? Both states relied on a self-selecting pool of parents who went out of their way to have their car seats inspected. It’s hard to say whether that group overrepresents people who know they’ve put the car seat in wrong, or whether it actually skews toward people who are super-conscientious about safety. I’m also not sure I believe that a state trooper who took a four-day course in car seat installation really has some magic technique that my wife and I were unable to divine in the many seat-strapping sessions we’ve taken part in since having a baby and acquiring a passenger car.
The standard for “correct installation” given in nearly every car seat manual in America is that the seat should no more than an inch in either direction “along the belt path” — i.e., from side to side. But because the seat is basically cinched down by a semi-circle of slippery nylon seatbelt, that standard is rank bullshit. I issue the following challenge to all car seat proponents everywhere: I can, with my arms of average strength, easily rock any car seat on the market, installed by whatever expert you choose, more than an inch along the belt path. And if I can do it, so can a Buick in a side-impact crash.
Which is not to say child safety seats are useless. There’s plenty of evidence that, for infants and young toddlers in particular, they’re quite effective at reducing injury and mortality in car accidents. But the “correct installation” phantom is representative of the way we think about child safety — we tend to want to make it into a measure of personal dedication, so that a parent who buys an expensive car seat and takes it to the technician to have it installed can feel like a “good” parent. But as this cranky mom points out, a system that lets people feel like they’ve won Safe Parent Of The Year is not necessarily a system that makes sense in terms of child safety:
It seems that auto manufacturer Volvo has partnered with carseat manufacturer Britax-Romer to create a custom-made carseat model that fits and works perfectly only in Volvo vehicles. The intention was to create a carseat that would be supremely safe when used in Volvo vehicles. The National Highway Traffic Safety Administration will not allow that carseat to be sold in the U.S. because the “NHTSA has mandated that child seats cannot be vehicle-specific”…”every child seat must fit in every car”….
[W]hat could be the NHTSA’s motive for denying the new product to be sold in the U.S.? Are they being irrationally bureaucratic? Are they trying to protect the consumer by not letting the market become flooded with vehicle-specific seats? Is inspection and regulation of so many seats too much of a logistical nightmare for all truly concerned with child safety? And wouldn’t the whole bloody mess just evaporate if all cars had the option to come pre-built with integrated safety seats?
Nonetheless, the debate over car seat safety occasionally produces interesting tangential results. Steven Levitt, the University of Chicago economist and co-author of the much-debated Freakonomics books, caused a stir in 2008 by claiming in the New York Times Magazine that for children over the age of two, seat belts are as effective as after-market car seats in reducing mortality and serious injuries. That’s a bold claim, and Transportation Secretary Ray LaHood attacked Levitt’s work as an attempt to be “provocative,” which it probably was. But he also claimed that Levitt’s work was based on a single set of data, the U.S. Department of Transportation’s Fatal Accident Reporting System, and that he used an overly broad timeframe in his study, reaching back as far as 1975, when there were no child car seats. (LaHood, not a scientist, derives these criticisms from a paper in the Archives of Pediatric and Adolescent Medicine which attempts to refute Levitt’s claims.)
It’s true that in Levitt’s early work on the subject he did rely exclusively on the FARS, and he acknowledges the limitations of his data set. (Levitt in fact goes further than his critics, noting that:
[I]n order for a crash to be included in the FARS data set, there must be at least one fatality. For those crashes in which a child is the only fatality, if that child would have survived the crash had he or she been differently restrained, the crash would have been absent from FARS. Similarly, for crashes in which no one dies, but a child would have died had he or she been using a different restraint, the sample composition depends upon the child’s restraint use. As demonstrated by Levitt and Porter (2001), this sample selection is likely to cause the benefits of effective safety devices to be systematically understated in the analysis presented above.
(You have to admire the cheek of a guy who cites his own studies when determining whether he has any blind spots in his data selection.)
But between 2005, when the first study was published, and 2009, when LaHood was complaining, Levitt had undertaken an additional study with Joseph Doyle, mining data from three additional databases: the General Estimates Survey, the New Jersey Department of Transportation’s figures, and the Wisconsin Crash Outcome Data Evaluation System (CODES). The newer study also restricted itself to more recent, car-seat-utilizing years. And the new study confirmed the results of the earlier study: seat belts are approximately as safe as after-market car seats for children over 2.
Levitt responded to LaHood’s criticism with all the social grace you might expect from a University of Chicago economics professor: he accused the Secretary of Transportation of not really caring about safety. Fair enough, since LaHood had said as much about Levitt, but what I found interesting were the comments on Levitt’s various blog posts on child seat safety. Some people, of course, reacted emotionally to the issue: this fellow celebrated the joy his son felt when his father illegally turned his seat forward on a lonely back road, while this anonymous commenter tastelessly brought Levitt’s dead child into the discussion. But other commenters, it seemed, were really inspired by the emotional intensity of discussing child safety to broaden their knowledge of science and to try to read and understand outside their comfort zone.
I find it heartening, in a weird way, that many non-scientists have waded into this discussion and tried to apply critical thinking to a technical issue in which they are not experts. For example, there’s this thoughtful comment arguing that we can’t test the effectiveness of car seats or seat belts under laboratory conditions (i.e., properly installed) — we have to test them as they’re actually used. And even this guy, who’s such a huge douchebag he brings up his law school class ranking (and misspells “Summa Cum Laude”) when someone wounds his pride, is at base defending his right as a layman to wrangle with science, to try to understand it and become better informed about its methodologies.
I find this invigorating. I’ve been a parent for just over six months, and already I’ve run into a lot of areas, like co-sleeping, where public health authorities make sweeping proclamations based on really sketchy and inadequate science. And I don’t necessarily blame them — science is often cautious and unsure, while pamphlets and billboards need to be bold, certain, and unwavering. But I do dream of a world that encourages a little more skepticism and curiosity and “What’s your source for that?”
(Elana has an idea that everyone should take a kind of personality test that would enable you to encode your own degree of scientific curiosity vs. your level of trust in authority. So if you’re a big fat jerk, like me, you could go to the doctor or the supermarket or wherever and say, “I’m a 2-F,” and they’d say, “Oh, all right. Here are the studies supporting this conclusion.” Or if you didn’t care and just wanted someone to tell you what to do, you could say, “I’m a 7-A,” and they’d say, “Here’s your pamphlet. It’s in bullet points!”)