I used to think that function tests should have been the official MBTI. In that view, E/I and J/P lose almost all of their significance as standalone factors, since they are just indicating the attitude of the functions. The 16 types might as well have been denoted in terms of the function-attitudes: "XeYi".
But then I saw how Jung really conceived of it more in terms of four functions (S, N, T, F), with the dominant paired with either one of the two attitudes, and from there, the other functions are also assigned attitudes. This is why classic MBTI theory only deals with "the first four" function-attitude combo's. The "other four" were just the rejected orientations of the first four.
This made it easier to understand several things. E/I retained their traditional meanings as far as behavior (no real "introverted extraverts"), and J/P also regain their meaning in observable behavior. The biggest thing understanding why the function-attitude strengths come out in all sorts of order in the function tests.
If your middle two letters seem settled, and it's just a matter of being close on J/P, then your preferred functions are likely sure; it's just a matter of which one bears the extraverted attitude. With the functions being measured, the suggested type can be all over the place if the "strengths" are to a certain extent out of order.