In so many cases, people are always stuck on one dichotomy; most commonly S/N or J/P. Why is this?
IMO: because the dichotomies are a weak and clumsy tool to determine someone's type.
I'm definitely an INFJ, and yet I systematically type as an INFP. Why? Because I really do like to keep my options open as long as possible before making a decision, for example. Things like that. Things that are supposedly "P". Except that, hey, my Dominant function is Ni, a Perceiving function! So *of course* I like to keep my options open: it's what Ni does... until it finally iNtuits some kind of solution.
The problem is that the dichotomy tests try to determine my *outside* behaviour (which is what P/J is about) by asking me how *I* see myself. Huh, bad idea, because as an Introvert, the way I see myself is mostly the way I am *inside*. If you want to know how I appear to others, go ask them, not me. If you ask me, I'm going to tell you how I see myself inside, because as an Introvert, that's my main field of interest. So instead of describing my Extraverted Auxiliairy, I'm going to describe my Introverted Dominant - hence the P conclusion.
The dichotomies could be an accurate way to determine someone's type IF they interacted with each other. "Ah, you're a strong Introvert, we need to take that into account when determining P/J". "Ah, you're a Feeler, we need to adjust the P/J questions to test for Fe rather than Te". Things like that. As they are now, the dichotomies are way too rigid and separate from each other to yield consistent results.
As just one tool in an array of tests, though, they can bring very valuable information. But only when taken along with other indicators (cognitive processes tests, temperament tests, interaction styles, and so on).