Disclaimer, first of all: I've had to reexamine and completely alter my arguments several times in this post, and I've moved stuff, edited stuff, and deleted stuff. I hope it makes sense and is at least somewhat logical.
It seems people constantly mix "Do you believe in God?" with "Does God exist?" Former is knowable, latter is not.
Using those two questions then, this is how I would define the religious terms:
A theist by definition would answer "yes" to both of those questions.
An atheist would answer "no" to both of those questions.
An agnostic would either answer "yes"
or "no" to the first, and "I don't know" to the second.
However, it should be noted that the first question is, under normal circumstances, the same as the second. The agnostic response provides the exception to this, of course, but if this were any other similar question the two you posed would be exactly the same. That is because by asking "Does god exist?" you are also asking "Do you believe in god?". With the 2 questions you gave, a sane person
cannot answer "yes" to one and "no" to another; they are either going to answer "yes" to both or "no" to both. It is only when the agnostic says "I don't know if a god exists" that you are required to ask "do you believe in god?" because answering the 2nd question with a yes or no automatically answers the first. When you ask someone if something exists, you are really asking them if they
believe it exists (the exception, again, being the agnostic, but I'm too tired to reason that out with logic). While the questions are logically different, I suppose they are interpreted to be the same due to social and practical conventions.
And second, the percentages don't make any sense. What's it like being 72% sure and how is it different from being 59% sure? Either you accept the claim or you don't.
Perhaps I made a mistake in that regard. As far as my mind can tell me right now, there is not a qualitative difference between 72% sure and 59% sure. The point I was trying to make (but perhaps did not emphasize well enough) is that there is a qualitative difference between 99% sure and 100% sure. If you were 100% sure, while you may still be lacking real and factual proof, your certainty is so strong that you truly think the proof is there (regardless of whether it truly is or not).