cjking, it doesn't make any difference. THAT is what I'm saying.
Look at the stats I posted a few above.
Death rates are 3-5x lower for minivans and large cars. So, overall, mass seems to trump maneuverability and design. That's not theory, that's not trying to derive actual safety from simulations with crash test dummies and brick walls, it's not based on a few anecdotes, they are real world facts.
For a variety of reasons, I prefer smaller cars. I was drawn to the Volvo S40 because a small well designed (for safety) car is batter than a small poorly designed one.
-ERD50
I agree that those stats are a useful basis of discussion, and render discussion about what proportion of accidents are head-on redundant.
However the way you have summarised them, by sorting, means you are in effect comparing the best cars with the worst. (For example, the best minivan versus the worst subcompact.)
I've computed the typical (median) death rate for each type.
If we make the very unrealistic assumption the we are altruistic enough to care about other peoples lives as much as our own, there's not much difference, with the death rate for subcompacts being 129 versus 114 for SUVs. Minivans total death rate is best at 72. I've looked at the original version of the table, and the others figure only includes the drivers of the other cars, not passengers, so these total figures can fairly be compared with each other.
Anyway, let's look at the stats for driver deaths, and forget about people other than ourselves we might kill. They do indeed show subcompacts as considerably worse than others. The figures are
Minivans 37
SUV 68
mid-size 70
compact 73
large 80
subcompact 95
pickups 111
I'm tempted to make something of pickups being worse than subcompacts, but I'll let it go.
Since subcompacts are only the second worse type, lets be symmetrical and for the sake of argument compare them with the second-best, SUVs at 68.
Your increased chance of dying driving a typical subcompact versus a typical SUV is 95-68 = 27. That's per million cars, so the probability is 0.0027%.
I assert (for the purposes of debate) that the average motorist implicitly values his life at one and a half million dollars, when making motoring purchase decisions. This figure is somewhat made up, it's something I half-remember from something I read a long time ago. If anyone is motivated to Google a better figure I'll be happy to revise my calculations. Note that the figure must relate to cars; people place different values on their life in different contexts. The place I remember this figure from made the point that people demand a much higher standard of safety on public transport - it seems we are willing to take more risks when we know we are in control as drivers, rather than depending on someone else.
The motorist who values his life at 1.5 million dollars should be willing to pay 0.0027%x1,500,000 = $41 more for the extra safety in a typical SUV, compared to a typical subcompact.
That's $41 per year the vehicle is driven, since the death rates are per year.
Remember my Prius versus Murano comparision, where the cost of fuel was about $6000 per year different, and you can see why I don't think safety will be a consideration when choosing my next vehicle.