What? You can compare the ratio of busses to cars to determine how many cars the bus removed from the road.

The number of cars removed from the road per bus is not dependent on the number of other cars on the road.

The number of cars removed is dependent on how many people ride in each bus and how many people ride in each car.

blackhaz didn't say it, but is probably thinking there are hundreds of cars per bus. My mental image was a bus with 10-20 people, vs. 500 cars (in that 20 minutes)

The logic still doesn't stand. If it's, say, 20 people, you shouldn't be comparing 1 car to 500 cars, you should be considering 1 bus to 20 cars.

One could argue every passenger on a bus saves some fraction of a car trip. Even if the total number of car trips saved is a much smaller number than the total number of cars, there's a 'breakeven' point above which the pain of the buses is less than the pain of the bus passengers in cars. I bet in most cities, we've hit that breakeven point, regardless of bus behavior.

<wild speculation> I bet that number is about 6 passengers per bus, excluding environmental concerns. </wild speculation>

Your wild speculation started earlier than your last line.

I think a good way to estimate it is to get a graph where you have average number of people in a bus at a time, then compare that to the average amount of people that are in a car at a time.

people_in_bus/average_#_people_in_cars = #_of_cars_removed_at_this_moment . Of course the model can be made significantly more complicated, but it's a good starting point.