In addition to the effects of car pollution on health, education, etc., car infrastructure is also a big driver of segregation in cities, which leads to a lot of the social problems that engender violence. So shifting infrastructure away from cars is really relevant.
A lot of freeways got built right in the middle of poor neighborhoods. I'm not saying there is a direct causal link with that and murders, but it's pretty hard to ignore the socio-economic effects of those massive infrastructure projects.
I think we generally underestimate the importance of social mixing in our society. Groups are less likely to see other groups as enemies if they're neighbors and take the same train to work every day. Poor people are less likely to feel excluded from society if they have friends from the middle class than if their whole neighborhood is on welfare.
Walkable neighborhoods with good infrastructure where a mixture of all strata of society live are, in my opinion, quite important if you want to prevent problems. Carcentric city planning is directly opposed to that.
Walkability isn't the issue in major cities, nor is proximity to other classes of people. In Baltimore (to stick with my original example), your likelihood of being a crime victim varies block to block, with one block being safe and the next being an open air drug market. That's not unusual for major cities in the US.
