This article was posted on one of my forums and people wanted a response to it.
The core of the article is that "Even a quick look at the FARS [Fatality Analysis Reporting System] data reveals a striking result: among children 2 and older, the death rate is no lower for those traveling in any kind of car seat than for those wearing seat belts." In other words, child safety seats and the laws requiring them at increasingly high ages provide no advantage over seat belts. The author states that even when the data is controlled for the severity of the crash and the size of the cars, it still holds that car seats and seat belts are roughly equivalent in protection.
He doesn't even ask a question that arises to my mind: does this data mean much if, as the author quoted near the beginning of the article, over 80% of carseats are improperly installed or used? If the vast majority of carseats aren't used properly -- aren't installed properly, the straps are positioned incorrectly for the child's size, straps left loose rather than tightened properly, etc -- then the vast majority of seats aren't going to be offering optimal protection. So the real world data might show only that improperly used seats don't offer additional protection over a seat belt.
The authors did look to test seats that were properly installed, and described in detail a crash test comparing properly used seats to the seatbelts alone. The result: the head and chest forces on the dummies were the same, rather they were in a carseat or only a seatbelt. Ah, but there is a huge caveat: "the sensors didn't measure neck or abdominal injuries, which child-safety advocates say are worse with seat belts." So the crash test didn't even measure the forces which car seat advocates say differ between child seat and seat belt use, exactly those forces which they say lead to child fatalities. Booster seat use, for instance, is advocated to prevent abdominal injuries from the lap belt, especially likely if the child puts the shoulder belt behind his back because it doesn't fit properly.
I would agree that many of the child restraint laws aren't very sensible. The issue doesn't have to do with the age or weight of the child much at all, but how well they fit (and stay) in a seat belt. And having adjustable seats and belts in cars (the authors' solution) makes sense. Using our built-in adjusters I've gotten an almost acceptable fit in our van with our nearly five year old. (He does ride in a booster seat). Under most laws, however, even if the seat belt could be adjusted to fit a five or six year old, it would not be legal to use it without a booster. And this varies so much from car to car -- in some cars the seats and belts can barely be adjusted to fit a small adult, let alone a 9 or 10 year old.