This may be good advice today, but will it hold five years from now? Ten years from now? Eventually, the delta in death rates per mile traveled will be indistinguishable from zero - either because safety improvements will no longer be significant, or because we have gotten the death rate so low that the difference between it and zero will be negligible.

We're currently at a point where 15-year old vehicles have ~144 deaths per million registered vehicles [1], and 5-year old vehicles have ~40 deaths per million registered vehicles [2]. Sure, you may be willing to pay a premium to get 40 down to 10, but are you willing to pay as much of a premium to get that 10 down to 2.5? 2.5 down to 0.6?

[1] https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/... (~2004 - 14.4 occupant deaths per 100,000 vehicles)

[2] https://www.iihs.org/ratings/driver-death-rates-by-make-and-... (~2016 - 14 to 64 driver deaths per 1,000,000 vehicles)

The demographics of those who drive 15 year old cars and those who drive 5 year old cars are very different. I'd bet that the 15 year old demo contains more reckless/intoxicated drivers (i.e. young people)

Yep, my wife rear-ended a delivery truck in our older Audi and totalled it -- with the kids in the car; everyone was OK, but after that she would not consider anything that didn't have collision avoidance systems and safety at least as good as the Audi we had. I'm glad she wasn't in our older Kia. There would have been injuries.

Exactly. Though I will say there are lots and lots of people who get cars they cannot afford or pay stupid money for ones, and there's way more of them then people who can afford more but choose the cheaper car.

Money well spent honestly. I've gotten in two accidents (one of them a t-bone) in cars made 2010ish and I thought they did fine both times with us walking away with minor bruising and a sore neck. But I can absolutely respect that.