Regardless of how this statistic is calculated, it seems like an incorrect one to be looking at, since it doesn’t address the absolute level of risk involved.
Epidemiology has some good definitions for dealing with the risk of a low probability event under various actions. I imagine a rationalist would find a statistic like “I would need to walk through a neighborhood of people not of my race 600 times for me to expect to be mugged 1 additional time” more likely to change his or her behavior.
Regardless of how this statistic is calculated, it seems like an incorrect one to be looking at, since it doesn’t address the absolute level of risk involved.
Epidemiology has some good definitions for dealing with the risk of a low probability event under various actions. I imagine a rationalist would find a statistic like “I would need to walk through a neighborhood of people not of my race 600 times for me to expect to be mugged 1 additional time” more likely to change his or her behavior.