surely you're not saying that you think car seats have NOT reduced child death rates in cars, are you?
I genuinely don't know. Have child fatalities over the past few years reduced faster than adult fatalities? Adults are dying on the roads at about half the rate they were a decade ago, and the drop has continued long, long after seat belts were made compulsory. Adult seatbelt compliance is high, and has been for a generation (you certainly see far fewer people on the roads not wearing seatbelts than you do children not correctly restrained). There are a huge slew of factors: cars are massively safer than they were both actively (ABS, stability control, modern tyres and suspension; compare with a Ford Anglia with crossply tyres, drum brakes, beam axles, friction dampers, etc) and passively (airbags, seatbelts, crumple zones, door bars, etc). But also rates of drink driving are lower, speed enforcement is better, young men are being priced off the road, the driving test has become massively harder than it was, road design has improved. Somewhere in there, something's reducing death tolls. Your PhD is assured if you can figure out what.
It's also interesting that there's been extensive research into risk homeostasis for seatbelts, but none for child seats. If people think they and their children are safer, do they drive more recklessly to "use up" the improved margin? There's some evidence that's true for seat belts, and anecdotal evidence that ABS has had a similar effect.
A related issue, cycling helmets, is incredibly nuanced, and what seems "obvious" is in fact nothing like obvious. Child seats definitely reduce the effect on a child of the accident once the accident has started. It's unlikely they do any harm. It's likely they do good. As to if they are overall the, or even a, reason why mortality rates on the roads are reducing is an excellent research topic.
FWIW, I had my kids in rear-facing seats until they were too old to use them, but I use trains wherever possible.