I've read that a spark plug makes a great auto window glass-breaker. I guess the theory is that it lets you combine just the right amount of force and speed on the right tiny section of the glass. Never tried it, though, nor saw it done.
You know, this is really a terminal ballistics question. One factor in breaking brittle things seems to be the speed with which the force of the striking object is transferred to the target. For example, try cradling an ice cube in the palm of your weak hand, and striking it swiftly with the bowl of a spoon held between the fingers of your dominant hand. You'll likely find that a teaspoon shatters the ice a lot better, most of the time, than a much-larger tablespoon. Try it a few times and see if it isn't true. I'm not sure of the physics behind this (Cliff, maybe you can elaborate), but I suspect it's got to do with all of the kinetic energy of the smaller spoon being transferred to the ice in a shorter timeframe, instead of a slower transfer that you get as the larger tablespoon decelerates over a slightly longer period of time. Maybe a similar phenomenon to how a .223 bullet may make a larger hole in a game animal than a .308.
Maybe the physicists on the forum can shed light on this?
Also, another potentially relevant question: is a glass-breaking technique that works when there's air on both sides of the glass going to work the same if there's water on the other side? I could see the glass behaving differently if the car's sinking in a river, than if you're testing it in a junkyard.