This morning’s NY Times has a couple of interesting stories. The first is that apparently the rate for bribery in NY State and City legislatures is as low as $20K. But the one that really caught my attention was the headline titled “Crosswalks in New York are not Havens.” Having three children who walk to school in Manhattan, this is a topic very near to my heart.
The article starts out with the following quote:
Pedestrians struck by cars are most often hit while in the crosswalk, with the signal on their side.
Pretty much every morning when I walk with my kids I make a point of saying that the signal is just a hint as to when it might be safe. What they really have to do is look for the cars. In fact, because the signal is just a hint it can to some degree be dangerous because it provides a false sense of safety!
But I of course also immediately thought of this being a classic Bayesian inference problem where we need to understand so-called base rates. Even in New York where jay walking is common, I would estimate that pedestrians crossing roads either against the signal or not at the crosswalk account for only about 20% of all crossings (probably even lower in reality). So unless doing so would be massively (at least 5x) more dangerous, we should expect for more people to be injured when they are doing the right thing.
Maybe it’s time again for me to teach my Skillshare class on Bayesian Probability and a Theory of Life. The goal for the class is for anyone who has taken it to question a headline and conclusion like this one about crosswalks.