They look at people who switch into entrepreneurship, and then consider the whole sample, so I don't think survivorship bias is a big issue.

I agree it would be good to go more fine grained in the categories though.