In the kinds of forces generated in a collision - especially a head-on which can have impact velocities north of 100mph pretty easily - small differences can play an outsized role. I suspect it’s not an either-or design question. We can design a restraint system that protects a wider range of body types, we just haven’t.
Some of it is cost, yes. The most effective crash mitigation tech involves seat inserts that are custom-molded to the individual’s body. That works well for astronauts, but would be horribly impractical for regular people. Especially since regular people tend to go up and down in weight, sometimes enough that they’re changing clothing sizes, and that would mean they need a new seat insert.
But some of it is simply inertia. We’ve always designed with this set of assumptions, and we aren’t going to stop now. While we aren’t going to be doing custom enclosures for drivers any time soon, we could redesign the idea of the seat belt such that it offers better protection. Maybe move up from the standard 3-point into 4? Yeah, you’d have to reinforce the roof for that, but that’s probably something that should be done regardless.
The low percentage thing is a bit of a red herring. We aren’t talking weight differences here - a system that can restrain a 200 pound weight can also restrain a 100 pound weight. We’re talking about weight distribution differences. If we designed a more universal restraint system then we could capture close to 100% of body types versus the 50-odd percent we’re capturing now.
But, again, and given some of the folks participating in this thread I feel it almost necessary to reiterate this with every post, it’s corporate laziness and apathy that’s preventing this research. It is not a car company conspiracy to kill or harm women.