An ongoing problem in nuclear physics is our ability to predict the rate of neutron and gamma emission of short-lived nuclei. This is especially true for beta-delayed neutron emission, an important but understudied reaction relevant to r-process nucleosynthesis and the decay of fission fragments. In this dissertation, I examine three hypotheses to explain this puzzling discrepancy between our models and experimental results. To improve our predictive capabilities, I have developed a suite of new computational tools which combine nuclear structure from the shell model with statistical reaction codes. Finally, I argue in favor of such a unified framework and I use the beta-delayed neutron emission of Rb-94 as a case study.