But what about the good old EPR paradoxon? The socalled "realistic" interpretations of the state, i.e., of the wave function or the state vector, representing a pure state, (or perhaps even the general case of a statistical opertor?) as a physical entity leads to the trouble with Einstein causality as detailed in EPR's famous paper.
So I think it's more save to follow the minimal statistical interpretation (or ensemble interpretation), according to which the state describes the probabilistic features of physically real systems (say an electron, just to avoid the additional quibbles with massless particles like photons), which can only be assessed by preparing ensembles of systems in the state to be investigated.
Formally, then not the abstract entities of the theory (vectors/statistical operators in Hilbert space representing the states and self-adjoint operators representing the observables) but equivalence classes of definite preparation procedures represent the "real system".