The fragmentation of molecules under conditions that result in yields of products that are thermodynamically controlled can be readily studied with statistical models. We explore which parameters influence the branching ratios using our recently developed Statistical Molecular Fragmentation model (SMF) and apply it to the decomposition of propane. We find that the fragmentation process has low sensitivity to the differences between the molecular descriptions given by commonly used ab initio methods (B3LYP, CCSD(T) and composite methods with different atom-centered basis sets). However, the branching ratios are most influenced by the vibrational frequencies of the molecules and radicals present in the decomposition pathways.