We recently proposed a concentration fluctuation model to describe the segmental dynamics of miscible polymer blends [Kumar , J. Chem. Phys. 105, 3777 (1996)]. This model assumes the existence of a cooperative volume, similar to that in the Adam-Gibbs picture of the glass transition, over which segments have to reorganize in a concerted fashion to facilitate stress relaxation. No molecular theory exists for the cooperative volume. Consequently, here we critically compare two alternative functional dependences for this quantity in the context of the segmental dynamics of the most extensively studied miscible polymer blend, 1,4-polyisoprene (PI) and polyvinylethylene (PVE): (a) The Donth model, which assumes the Vogel form for the temperature dependence of relaxation processes, with a relaxation time that diverges at the Vogel temperature, roughly 50 K below the glass transition, and (b) a more recent dynamic scaling model that predicts the relaxation time diverges algebraically, only about 10 K below the glass transition. We find that the dynamic scaling model provides a near-quantitative description of the segmental relaxation in PI/PVE blends. In contrast, the Donth model predicts that the relaxation time spectrum for PI, the faster relaxing component, is bimodal, in qualitative disagreement with NMR experiments and our dielectric measurements reported here. Our results therefore emphasize two findings. First, our model can describe the segmental relaxations of the components of a polymer blend in a near-quantitative manner. Second, and more fundamentally, it appears that the dynamic scaling model describes segmental dynamics of polymers near their glass transition.