Good question. I think in theory no. In practice it could look like this-for several reasons:Some people develop an unhealthy fixation with their dominant function, and this can create the impression of being more A. This is because the person is allowing function A to be gratified to excess. Another example is that some people may place more of an emphasis on one of their other functions-for instance they may call upon a weaker N because there job is to work in theoretical physics. So on that note it's possible sub types are really just a way of expressing such a functional preference (although more research on sub types is required).
So what this means is that personal emphasis can change but the actual type would remain static.
I think that % of A can be useful in testing-if someone is close to 50% then they need to spend more time investigating to determine their preference, and if it comes up as 100% then that's a suggestion that the person is wise to the test and the questions it asks. So I wouldn't take this percentage on a populace base, but it might be useful to use it representing only the questions the test itself asks, simply for the validity of the result.