Let me clarify things a bit. Insurance companies do correlation studies to create statistical models so they can predict risk factors and adjust premium plans. But, statisitics can be used to prove anything - 40% of all people know that.

The study I cited was an internal study to improve their premium models in order to maximize profits, minimize losses, and out perform the competition. So, you'll probably never see car color cited as the fault for an accident. It just seems like a weak, far fetched argument.

In my opinion, there are no insurance friendly colors because 1 benefit will have a complimentary disadvantage. In the case of red and yellow, you may be less likely to be struck by an inattentive driver, but you're more likely to get pulled over. If this is true (I'm not 100% certain), problems associated w/ car color should average out in the long run, and thus have neglible effect on your premium. So, get whatever color you like best.