I know what you're saying, and again, its very simplified. "top
heavyness" is to the rollability of cars on an impact, what a tennis
ball hitting a wall is to two cars impacting. its a very simplified
and overly basic evaluation.

I'll put it simply, into moresimple terms. a heavier car carries more
energy.
A taller car has a higher center of gravity.

now, in an impact between a large car and a smaller one, the intial
ofrces may be ---> <---
If, however, the surfeaces are not completely perpendicular, and the
collapsing equally uniform, there will be a vertical element to it, as
a proportion of its total energy. I greater vertical energy value (due
tot he greater overall energy value) means a greater rotational moment
about the pivot. The higher the COg, the less of a rotation the
vehicle needs to roll. So, a larger vehicle has a greater energy, and
a higher COG meaning its rollover will be greater than a smaller
vehicle for a given deflection.

Next, add in the physical height. A bigger vehicle has bigger wheels.
Often much greater suspension travel. Its impact sports will impact
higher up on the other vehicle, and thus will ride over. This produces
a greater deflection value for a bigger vehicle.

Bigger deflection, rotating with more energy, and a higher COG = very
liekly to roll in an impact with a smaller vehicle.

Its not rocket science. you just ca't look at it in terms of wooden
blocks for vehciles. you have to look at shapes, energies, and a
step-by-step look throughout the collision. Its like you're watching a
ball hit a baseball bat and thinking ball (small light) hits bat
(bigger heavier) so ball will always come off worst (and go out of the
park) It aint so, its curved, there's all sorts of energy deflections
and transfers. A simplistic model, such as the one you're using is
fine for a very eough estimate. but thats all it is, a very simplistic
ball-park figure. Heck, you rear-end someone, a big factor is their
brakes. if they were applied, a lot more damage is done thn if they're
not, yet your model takes none of that into consideration.

just as a closer, another video clip springs to mind - its of a
security post arangement. one of those powered bollards. Whole thing
was probably about 300lbs all told, for the enetire system. They
tested it by driving a 5ton truck into it at 30mph. Stopped the truck
dead. by your modeling methods, the post should probably have been
napped right off. It was, after all, a lot smaller and lighter. Truck
stopped, bollard was through to between the drivers seat, and the
truck almost ended up on its nose (high cog again) but it stopped it.

In short, its not so much the magnitudes of the energy that make the
difference, its where those energies go.