On the statistical derivation of the Schrödinger equation