That makes sense. It still seems to be more of a rhetorical tool to illustrate that there is a spectrum of subjective belief. People tend to lump important distinctions like these together: "all atheists think they know for certain there isn't a god" or "all theists are foaming at the mouth and have absolute conviction", so for a popular book it's probably a good idea to come up with this sort of scale like this, to encourage people to refine their categorization process. I kind of doubt that he meant it to be used as a tool for inferring Bayesian confidence (in particular, I doubt 6.9 out of 7 is meant to be fungible with P(god exists) = .01428).