The analysis of the transverse magnetization decay is a well-established method to obtain information about network parameters of elastomers or polymeric melts. The starting point is the scaling concept introduced by Cohen-Addad, which reduces the detailed description of the atomic bond-vector motion to that of a larger scaled subchain motion. When considering polymer networks some simplifications in the calculation of the NMR response are widely used. In the frozen bond assumption all the crosslink positions in a network are taken to be at fixed points, with the intercrosslink network end-to-end vectors having a Gaussian distribution. In the second moment approximation it is assumed that there is a Gaussian distribution of dipolar interactions, and additionally an exponential correlation function of the motion is used. Both models are able to explain the non-exponential magnetization decay (FID of a Hahn-echo NMR experiment). We will compare these different starting points to give some relations between them. Both models are tested by a NMR-relaxation experiment.