Ideally you would be able to carry out such an experiment but as troystribling mentions it's possible that an increase in the safety of cars may lead people to

Ideally you would be able to carry out such an experiment but as troystribling mentions it's possible that an increase in the safety of cars may lead people to drive more dangerously. It's hard to isolate this correlation. An experiment may be to show that holding everything else constant, a car from 2009 would lead to 50% fewer serious accidents. If we look at the data and notice that there were only 25% fewer accidents, we would then conclude that it may have come from the factors mentioned.