http://www.theboxotruth.com/
TONS of little experiments they do. Look for The Buick Of Truth on there somewhere.
I've seen that one, love the stuff he does. We've shot up a number of different vehicles, and found similar results, although cars with more severely sloped windshilds tend to actually deflect the bullets upward, leaving a long gouge before acheiving penetration. Rifle rounds, OTOH, tended to pass through very cleanly with minimal deflection.
While these static tests paint a grim picture for the occupants of a vehicle, real-world dynamic shootings involving cars tend to produce a great many spent casings with few or no injuries. He cites that the side glass isn't much of an obstacle, which is true.......when it's struck at a perpendicular angle. But as with any hard surface, change that angle and the results change with it. We found that many bullets fired from angles of < 30* didn't actually enter the vehicle at all, even though the glass shattered.
He also touched on the fact that there are a lot of heavier components inside the doors, which they managed to missed. Well, we've done this with more vehicles, more guns and A LOT more ammo, and found that door impacts frequently do not achieve penetration into the passenger compartment, especially with handguns and small rifle rounds. Window motors are really effective bullet stoppers if you happen to hit them, and cars that have steel regulators; the steel of those arms is pretty heavy, between 14 and 10 gauge. Heavy sheetmetal like that will stop most service handgun calibers.
While I certainly wouldn't want to be in a vehicle that was getting shot up, I also wouldn't want to be trying to defend against someone who was in a car. There are just so many variables that it's virtually impossible to predict what the bullet will do and where it will end up, especially in a dynamic situation.