> While it may be possible that an improperly installed car seat may not be any more effective than a seat belt alone, and that probably accounts for at least some of the Freakanomics data, that is a SOLVABLE problem.

That's not going to explain the data. The numbers require seat belts to be about as safe as car seats for children 2+ on average. Thus, unless an improperly installed car seat was significantly worse than an average seat belt, it's all kind of a wash.

Anyway, this is more about public policy than individual choice. If your spending public money to promote something it must increase the average safety levels or your wasting peoples money.

PS: As to improper use, when swapping multiple car seats it’s very easy to forget to attach one. With seat belts it’s more visually obvious their not in use. That’s a huge safety risk and could account for the unexpected data.