Can anyone speak to the reliability of using metropolitan statistical areas for something like this? Having lived across on both sides of the tracks in a few, grouping them for something like this seems like an interesting choice. One that I probably wouldn't agree with, but I'm out of my depth