I could use that same logic to refute a study showing public transportation is cost-effective as long as that study wasn't done in San Francisco. It would be an obviously incorrect argument in that case, and that is also true in your case regarding cars.

The study's numbers are far more connected to reality than your made up numbers. You are making an isolated demand for rigor.[1]

1. https://slatestarcodex.com/2014/08/14/beware-isolated-demand...

I'm not making an isolated demand for rigor, because I'm demanding everything we do does not include clear systematic biases in one direction. I'm making a global demand for the most minimal amount of rigor you could possibly use and not necessarily arrive at incorrect answers.

Your study has a clear systematic bias (edit: Actually it doesn't even say what you claimed it did, but if it did it would have a clear systematic bias), that most costs scale with population density, and it was done studying an area (the US) with a much lower population density than the area at hand (manhattan). Assuming your study is entirely correct, its answer is incorrect when applied to the problem at hand. You would have to be an idiot to rely on it because the only way it is "right" is by being incorrect by the right amount in the right direction by sheer chance.

Likewise, if you tried to apply a study about public transportation done in the middle of the Sahara Desert to San Francisco, I would call you an idiot.