Here's my confusion. Say if I went to some n-dim representation of the defining rep of $SO(3)$. would the wavefunction be represented by $\psi(V)$ where $V$ is $n$ dimensional. I would then need to know the map to get from $V$ to the normal $X$ right so I actually knew what was happening in real life.
However, people just arbritarily change the Dirac spinor to something else. How is this allowed? Surely this will change all expectation values?