Where I'm from, drivers are sitting in sound proofed boxes called cars. The newer the car, the better the sound isolation. So unless the taxi was a convertible, the comparison is invalid.
When I'm wearing my headphones I can easily understand someone talking to me. When I'm sitting in my car I don't hear a word when someone is talking outside the door.
Are these the values from one specific year? If yes, the statistical error would be massive, especially for the cities <50k. The value of 73/100k in Dayton means, that ~7 people died, which could easily be caused by a single incident. So it might be the case that Dayton is generally a safe city, but it was just a really bad year for them. You'd have to look at multiple years to be able to see a pattern.