A major potential difference between polymer and small molecule organic light-emitting devices is the possibility of spin-dependent charge recombination in the latter. In this case, the singlet exciton formation yield, one of the key parameters that determines device efficiencies, deviates from the simple quantum mechanical spin statistics prediction of 25%. For polymers, the departure from this limit has been demonstrated by several indirect measurements yielding a wide spread values from 25% up to 95% and by ourselves using a direct measurement on working devices at 44%. These results clearly indicate that polymer-based devices hold a unique advantage over small molecule-based devices. Comment will be made to these various measurement techniques along with an insight into current theoretical ideas that can explain the departure from the quantum spin statistical limit.